A master micro-grid collaborative optimization scheduling method

Abstract

The application relates to the technical field of power system dispatching, in particular to a main-distribution-micro-grid collaborative optimization dispatching method, which solves the technical problem that it is difficult to realize the main-distribution-micro-grid collaborative optimization dispatching under the voltage safety constraint of a distribution network. The method comprises the following steps: performing coordinate rotation on a power change amount based on a power control trajectory angle to obtain a control-axis power component and an off-axis power component; determining the active voltage sensitivity of each micro-grid to each monitoring node based on the control-axis power component, the off-axis power component and the differential voltage response of the voltage of each monitoring node relative to the gateway voltage; constructing a voltage safety constraint boundary based on the active voltage sensitivity, the current voltage of each monitoring node and a preset voltage safety threshold; and solving a preset power distribution optimization model by taking the total power dispatching instruction and the voltage safety constraint boundary as constraints to obtain the active power set value of each micro-grid.

Description

Technical Field

[0001] This application relates to the field of power system dispatching technology, specifically to a method for coordinated optimization dispatching of main and distribution microgrids. Background Technology

[0002] As the penetration rate of distributed power sources such as distributed photovoltaic and energy storage systems in distribution networks continues to increase, distribution networks are transforming from traditional unidirectional power receiving networks to multi-source interactive active distribution networks. In the coordinated dispatch of the upper-level power grid and the distribution network, the distribution network needs to aggregate the regulation capabilities of multiple distributed power sources within it to respond to the total power dispatch instructions issued by the upper-level power grid and achieve coordinated and optimized operation of power sources, grid, load and storage.

[0003] In the existing technology of coordinated dispatching of main and distribution microgrids, the distribution network needs to rationally allocate the power regulation tasks of each microgrid based on the impact of microgrid power regulation on node voltage, i.e., voltage sensitivity, to avoid the problem of node voltage exceeding limits during dispatching. However, in practical applications, existing dispatching methods are difficult to accurately obtain effective voltage sensitivity parameters, resulting in poor aggregation effect of the distribution network on the microgrid's regulation capabilities. When responding to dispatching commands from the main grid, this can easily lead to voltage exceeding limits at key nodes in the distribution network, causing inverter protection to disconnect from the grid. This seriously affects the safety and reliability of coordinated dispatching of main and distribution microgrids and cannot meet the safety dispatching requirements of the distribution network under high-penetration distributed power source access. Therefore, how to achieve coordinated optimization dispatching of main and distribution microgrids under distribution network voltage safety constraints under the conditions of lacking line impedance parameters and the existence of multicollinearity in the data has become an urgent technical problem to be solved. Summary of the Invention

[0004] To address the technical challenge of achieving coordinated and optimized scheduling of main and distribution microgrids under voltage security constraints in distribution networks, this application aims to provide a method for coordinated and optimized scheduling of main and distribution microgrids. The specific technical solution adopted is as follows: Obtain the gate voltage of the upstream power grid connected to the distribution network, the voltage of each monitoring node in the distribution network, and the active and reactive power of each microgrid connection point in the distribution network; Based on the power changes in active and reactive power, the power control trajectory angle of each microgrid is determined, and the coordinates of the power changes are rotated based on the power control trajectory angle to obtain the control axis power component and the deviation axis power component. Based on the control axis power component, the off-axis power component, and the differential voltage response of each monitoring node voltage relative to the cut-off voltage, the active voltage sensitivity of each microgrid to each monitoring node is determined. Based on the active voltage sensitivity, the current voltage of each monitoring node, and the preset voltage safety threshold, a voltage safety constraint boundary is constructed. In response to receiving the total power dispatch command from the upper-level power grid, the preset power allocation optimization model is solved using the total power dispatch command and voltage security constraint boundary as constraints to obtain the active power setpoint of each microgrid.

[0005] In one possible implementation, the power control trajectory angle of each microgrid is determined based on the power changes in active and reactive power. This includes: performing linearity pre-verification on the active and reactive power change data of a single microgrid within a preset historical time window; determining the power control trajectory angle based on the data within the historical time window when the correlation between the active and reactive power change data meets the preset linearity condition; and using the power control trajectory angle from the previous moment as the power control trajectory angle for the current moment when the correlation does not meet the preset linearity condition.

[0006] In one possible implementation, the power control trajectory angle is determined based on data within a historical time window, including: summing the active power change data within the historical time window to obtain an active power sum; summing the reactive power change data within the historical time window to obtain a reactive power sum; using the active power sum as a component of the active power coordinate axis and the reactive power sum as a component of the reactive power coordinate axis to form a composite vector; calculating the angle between the composite vector and the positive direction of the active power coordinate axis using a preset angle determination algorithm, and using the angle as the power control trajectory angle to characterize the dominant power change direction of the microgrid.

[0007] In one possible implementation, before determining the active voltage sensitivity of each microgrid to each monitoring node, the method further includes: constructing an input feature vector for the current moment based on the control axis power components and off-axis power components of all microgrids at the current moment; calculating the norm of the input feature vector and comparing the norm with a preset excitation dead zone threshold; when the norm is less than the preset excitation dead zone threshold, determining that the current moment is in an identification silence period, and keeping the active voltage sensitivity determined in the previous moment unchanged; when the norm is greater than or equal to the preset excitation dead zone threshold, determining that there is effective excitation at the current moment, and determining the active voltage sensitivity of each microgrid to each monitoring node.

[0008] In one possible implementation, determining the active voltage sensitivity of each microgrid to each monitoring node includes: using the differential voltage response as the identification target, the control axis power component and the off-axis power component as the identification input, and using a recursive estimation algorithm with a forgetting factor and regularization coefficient to iteratively update the axial sensitivity coefficient; and performing an inverse transformation to restore the updated axial sensitivity coefficient and power control trajectory angle to determine the active voltage sensitivity.

[0009] In one possible implementation, a voltage safety constraint boundary is constructed based on the active voltage sensitivity, the current voltage of each monitoring node, and a preset voltage safety threshold. This includes: for any monitoring node, constructing a linear inequality constraint on the power setpoint of each microgrid based on the current voltage of the monitoring node, the preset voltage safety threshold, the preset voltage safety margin, and the active voltage sensitivity of each microgrid to the monitoring node; and using the combination of the linear inequality constraints corresponding to all monitoring nodes as the voltage safety constraint boundary.

[0010] In one possible implementation, a preset power allocation optimization model is solved to obtain the active power setpoint for each microgrid. This includes: establishing a power allocation optimization model with the objective functions of minimizing the power regulation amplitude of each microgrid and minimizing the deviation between the power of each microgrid and its preset economic optimal operating point; using the total power dispatch command as an equality constraint, and the voltage safety constraint boundary, the physical output limit constraint of each microgrid, and the trust domain step size constraint used to limit the single power regulation amplitude as inequality constraints, which together constitute the solution conditions for the power allocation optimization model; and solving the power allocation optimization model that satisfies the solution conditions to obtain the active power setpoint for each microgrid.

[0011] In one possible implementation, the process of determining the power change includes: calculating the time difference components of the voltage at each monitoring node, the time difference components of the threshold voltage, and the time difference components of active power and reactive power respectively; and using the time difference components of active power and reactive power as the power change.

[0012] In one possible implementation, the process of determining the differential voltage response includes: for any monitoring node, performing a difference calculation between the timing difference component of the voltage at the monitoring node and the timing difference component of the gate voltage, and using the calculation result as the differential voltage response corresponding to the monitoring node. The differential voltage response is used to characterize the relative voltage change caused by power regulation within the distribution network.

[0013] In one possible implementation, the acquisition of the threshold voltage of the upstream power grid connected to the distribution network, the voltage of each monitoring node in the distribution network, and the active and reactive power of each microgrid connection point in the distribution network includes: using a preset scheduling cycle as the time reference, collecting the current threshold voltage amplitude, the voltage amplitude of each monitoring node, and the voltage amplitude, active power value, and reactive power value at each microgrid connection point; when the timestamp of any collected data deviates from the current time, a linear interpolation method is used to synchronize all collected data to the same time segment.

[0014] This application offers the following advantages: By acquiring the threshold voltage, monitoring node voltage, and the active and reactive power at the microgrid's grid connection point, a complete foundation for distribution network operation status perception is established. By determining the power control trajectory angle and performing coordinate rotation, the physically strongly coupled active and reactive power are transformed into statistically decoupled control axis and deviation axis components, effectively solving the identification difficulties caused by data multicollinearity. By estimating the decoupled power components and differential voltage response, the active voltage sensitivity can still be stably output even under data rank deficiency conditions, achieving online identification of voltage influence patterns independent of line impedance parameters. By constructing a voltage safety constraint boundary based on the identified active voltage sensitivity and introducing this boundary constraint into the power allocation optimization model, the risk of voltage exceeding limits can be automatically avoided when responding to the overall power dispatch command from the upper-level grid. Based on this, this application can achieve coordinated optimization dispatch of the main distribution microgrid under distribution network voltage safety constraints under conditions of lacking line impedance parameters and data multicollinearity, significantly improving the distribution network's capacity to accommodate distributed power sources and its dispatch security. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a primary and secondary microgrid collaborative optimization scheduling method provided in one embodiment of this application; Figure 2 This is a schematic diagram of the system architecture of a primary and secondary microgrid collaborative optimization scheduling system provided in one embodiment of this application. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a primary and secondary microgrid collaborative optimization scheduling method proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0019] The following description, in conjunction with the accompanying drawings, details a specific scheme for a collaborative optimization scheduling method for main and distribution microgrids provided in this application.

[0020] Please see Figure 1 It illustrates a flowchart of a primary and secondary microgrid collaborative optimization scheduling method provided in one embodiment of this application, as follows: Figure 1 As shown, the method includes the following steps: Step 101: Obtain the gate voltage of the upstream power grid connected to the distribution network, the voltage of each monitoring node in the distribution network, and the active and reactive power of each microgrid connection point in the distribution network.

[0021] Among them, the threshold voltage refers to the voltage amplitude at the common connection point between the distribution network and the upper-level main grid or at the substation busbar. This data reflects the voltage background level of the upper-level grid and serves as a benchmark reference for subsequent differential calculations. Monitoring node voltage refers to the voltage amplitude of critical nodes within the distribution network that require voltage safety management. These nodes are typically located at the end of the distribution network or in electrically sensitive locations, and are key monitoring targets for voltage exceedance risks. The active and reactive power at the microgrid connection point refers to the active and reactive power values ​​injected into the distribution network by each microgrid, reflecting the power injection from the microgrid into the distribution network.

[0022] For example, this step uses a Supervisory Control and Data Acquisition (SCADA) system or an Advanced Metering Infrastructure (AMI) as a data aggregation platform. Using a preset scheduling cycle as the time base, it collects the current voltage amplitude at the control points, the voltage amplitude at each monitoring node, and the voltage amplitude, active power, and reactive power at each microgrid connection point. The scheduling cycle can be configured within the range of 1 to 15 minutes based on system response requirements and equipment sampling capabilities; for example, it can be set to 15 minutes.

[0023] In some embodiments, considering that the sampling clocks of different measuring devices in the distribution network may have slight deviations, when the timestamp of any collected data deviates from the current time, linear interpolation is used to synchronize all collected data to the same time segment. As an example, for a certain physical quantity, if its time... and The measured values ​​are respectively and ,and < < Then at time The synchronization value is calculated using linear interpolation and satisfies the following formula: .

[0024] Based on the above time synchronization processing, it can be ensured that all physical quantities involved in the calculation are attributed to the same physical cross section, providing an accurate data foundation for subsequent causal analysis.

[0025] 102. Based on the power changes of active and reactive power, determine the power control trajectory angle of each microgrid, and perform coordinate rotation on the power changes based on the power control trajectory angle to obtain the control axis power component and the deviation axis power component.

[0026] The power change refers to the difference between the power value at the current moment and the previous moment, reflecting the dynamic adjustment of power. The power control trajectory angle characterizes the dominant direction of power change in the microgrid on the active-reactive plane, physically reflecting the current power factor setting or control slope of the microgrid. The control axis power component refers to the component of the power change projected onto the power control trajectory direction, containing most of the energy fluctuation information. The off-axis power component refers to the component of the power change projected onto the direction perpendicular to the power control trajectory, mainly reflecting measurement noise or minor control jitter.

[0027] Optionally, this step first calculates the time-series difference components of active and reactive power in each microgrid as power variation, then determines the power control trajectory angle based on the power variation data within the historical time window, and finally decomposes the power variation into control axis components and deviation axis components through rotation transformation. This transformation converts physically strongly coupled active and reactive power into statistically decoupled characteristic variables, fundamentally solving the identification difficulties caused by data multicollinearity and providing a geometric basis for subsequent regularized recursive estimation.

[0028] Step 103: Based on the control axis power component, the off-axis power component, and the differential voltage response of each monitoring node voltage relative to the cut-off voltage, determine the active voltage sensitivity of each microgrid to each monitoring node.

[0029] Differential voltage response refers to the difference between the voltage at each monitoring node and the threshold voltage. It is used to suppress the common-mode voltage trend transmitted from the upper-level power grid, ensuring that the extracted voltage increment mainly reflects the relative impact of power regulation within the distribution network. Active voltage sensitivity characterizes the voltage change at monitoring nodes caused by a unit change in active power and is a core parameter for constructing voltage safety constraint boundaries.

[0030] In some embodiments, this step first calculates the differential voltage response of each monitoring node voltage relative to the cutoff voltage. Then, using this differential voltage response as the identification target and the control axis power component and off-axis power component as identification inputs, a recursive estimation algorithm with a forgetting factor and regularization coefficients is used to iteratively update the axial sensitivity coefficient. Since the off-axis component mainly contains noise and has weak excitation, the regularization term can effectively suppress the parameter estimates in the off-axis direction, preventing calculation divergence caused by matrix singularities. Finally, the updated axial sensitivity coefficient and power control trajectory angle are inversely transformed to obtain the active voltage sensitivity in the physical coordinate system. This mechanism enables the stable output of physically meaningful sensitivity parameters even under a single excitation mode where active and reactive power change proportionally, without requiring additional artificial disturbance experiments on the microgrid.

[0031] Optionally, the regularized recursive estimation algorithm refers to introducing a Tikhonov regularization term into the recursive least squares algorithm, which ensures the numerical stability of the algorithm under the condition of data rank deficiency by penalizing the parameter norm.

[0032] Step 104: Based on the active voltage sensitivity, the current voltage of each monitoring node, and the preset voltage safety threshold, construct the voltage safety constraint boundary.

[0033] The preset voltage safety threshold refers to the upper limit of voltage allowed by the distribution network, for example, set to 1.05 times the nominal voltage. The voltage safety constraint boundary refers to the feasible domain boundary formed by the power settings of each microgrid under the condition of satisfying the voltage safety constraint.

[0034] In some embodiments, for any monitoring node, a linear inequality constraint on the power setpoint of each microgrid is constructed based on the current voltage of the monitoring node, a preset voltage safety threshold, a preset voltage safety margin, and the active voltage sensitivity of each microgrid to the monitoring node. The combination of the linear inequality constraints corresponding to all monitoring nodes is used as the voltage safety constraint boundary, which is mathematically represented as a multidimensional convex polyhedron, limiting the power regulation range of each microgrid.

[0035] Optionally, the linear inequality constraint takes the form of: the sum of the current voltage and the voltage changes caused by the power regulation of each microgrid is less than or equal to the preset voltage safety threshold minus the safety margin. By introducing the safety margin, a buffer space is reserved to offset the uncertainty caused by identification errors and linearization approximation, ensuring that even if there are certain prediction errors, the scheduling result can fall within the absolute safety range with a high probability.

[0036] Step 105: In response to receiving the total power dispatch command from the upper-level power grid, the preset power allocation optimization model is solved using the total power dispatch command and voltage security constraint boundary as constraints to obtain the active power setpoint for each microgrid.

[0037] Among them, the total power dispatch instruction from the upper-level power grid refers to the target value of total active power dispatch for the distribution network issued by the main grid at the next moment. The preset power allocation optimization model refers to an optimization model that aims to minimize the power regulation amplitude and minimize the deviation between power and the economically optimal operating point, under the premise of satisfying the total power balance constraint and voltage safety constraint.

[0038] As one possible implementation, this step establishes a power allocation optimization model with the objective functions of minimizing the power regulation amplitude of each microgrid and minimizing the deviation between the power of each microgrid and its preset economic optimal operating point. The total power dispatch command is used as an equality constraint, while voltage safety constraints, physical output limit constraints for each microgrid, and trust domain step size constraints (used to limit the amplitude of a single power regulation) are used as inequality constraints, collectively constituting the solution conditions for the power allocation optimization model. The trust domain step size constraint limits the allowable power change amplitude in a single dispatch, ensuring that the local linearity assumption of the sensitivity parameter still holds and preventing the optimizer from searching for solutions far from the current operating point. The power allocation optimization model satisfying the solution conditions is solved by calling a standard quadratic programming solver to obtain the active power setpoints for each microgrid. Under this scheduling mechanism, active voltage sensitivity plays an automatic weight adjustment role. If a microgrid has a significant impact on the voltage of critical nodes, when the node voltage approaches the upper limit, the constraints will tighten rapidly, forcing the optimizer to reduce the power allocation of that microgrid and instead call on other microgrids with lower sensitivity to meet the total power command. This achieves active voltage limit avoidance based entirely on data.

[0039] Based on the above technical solutions, this embodiment establishes a complete foundation for distribution network operation status perception by acquiring the threshold voltage, monitoring node voltage, and active and reactive power at the microgrid connection point. By determining the power control trajectory angle and performing coordinate rotation, the physically strongly coupled active and reactive power are transformed into statistically decoupled control axis components and deviation axis components, effectively solving the identification difficulties caused by data multicollinearity. By estimating the decoupled power components and differential voltage response, the active voltage sensitivity can still be stably output under data rank deficiency conditions, realizing online identification of voltage influence laws independent of line impedance parameters. By constructing a voltage safety constraint boundary based on the identified active voltage sensitivity and introducing this boundary constraint into the power allocation optimization model, the voltage limit risk can be automatically avoided when responding to the total power dispatch command of the upper-level power grid. Based on this, this embodiment can achieve coordinated optimization dispatch of the main distribution microgrid under distribution network voltage safety constraints under the conditions of lacking line impedance parameters and data multicollinearity, significantly improving the distribution network's acceptance capacity for distributed power sources and dispatch security.

[0040] In one possible implementation, the process of obtaining the gate voltage of the upstream power grid connected to the distribution network, the voltage of each monitoring node in the distribution network, and the active and reactive power of each microgrid connection point in the distribution network in step 101 specifically includes: Step 201: Using the preset scheduling cycle as the time reference, collect the current voltage amplitude at the gate, the voltage amplitude at each monitoring node, and the voltage amplitude, active power value, and reactive power value at each microgrid grid connection point.

[0041] Specifically, this application can use a Supervisory Control and Data Acquisition (SCADA) system or an Advanced Measurement System (AMI) for the power distribution network as the data aggregation platform. The system sampling and scheduling cycle is set as follows: (In this embodiment, we take) =15 minutes, but in practice, it can be configured within the range of 1-15 minutes depending on the equipment capacity. (This refers to the current scheduling time.) Collect the following data: The voltage amplitude at the point of common coupling (PCC) between the distribution network and the upstream main grid or at the substation busbar is collected. This data reflects the background voltage level of the upstream power grid.

[0042] For each microgrid (i=1,2,...N) within the distribution network area, collect the voltage amplitude at its grid connection point. Active power injected into the distribution network and reactive power .

[0043] For each critical monitoring node (m=1,2,...M) requiring voltage safety control, the voltage amplitude of the monitoring node is collected. .

[0044] Step 202: When the timestamp of any collected data deviates from the current time, use linear interpolation to synchronize all collected data to the same time segment.

[0045] Specifically, after obtaining the above data, if it is found that the timestamps of each data point do not match the standard time... If the alignment is not perfect (e.g., the deviation exceeds 1 second), linear interpolation is used for synchronization. Assume a physical quantity x... and The measured value at time and ,and < < ,but The synchronization value at any given time is calculated as follows: .

[0046] The above process ensures that all physical quantities involved in the calculation are attributed to the same physical cross-section.

[0047] In one possible implementation, the process of determining the power change includes: calculating the time difference components of the voltage at each monitoring node, the time difference components of the threshold voltage, and the time difference components of active power and reactive power respectively; and using the time difference components of active power and reactive power as the power change.

[0048] The process of determining the differential voltage response includes: for any monitoring node, performing a difference calculation between the timing difference component of the voltage at the monitoring node and the timing difference component of the gate voltage, and using the calculation result as the differential voltage response corresponding to the monitoring node. The differential voltage response is used to characterize the relative voltage change caused by power regulation within the distribution network.

[0049] As an example, for time Utilize the current moment Data compared to the previous moment The time difference components of each physical quantity are calculated from the data: timing difference component of gate voltage Satisfying the formula: .

[0050] Timing difference component of active power in microgrids Satisfying the formula: = .

[0051] Time-series difference component of reactive power in microgrids Satisfying the formula: = .

[0052] Monitoring node voltage timing difference components Satisfy the following formula: = .

[0053] Differential voltage response Satisfy the following formula: = - .

[0054] In one possible implementation, the specific process of determining the power control trajectory angle of each microgrid based on the power changes in active and reactive power in step 102 above includes: Step 301: Perform linearity pre-verification on the active power change data and reactive power change data of a single microgrid within a preset historical time window.

[0055] Specifically, for each microgrid, a preset historical time window of length w is established (in this embodiment, it is recommended to use 12 to 24 sampling points, corresponding to 3 to 6 hours of operating data, to cover a sufficient range of light variation). The time series data within the window is extracted: .

[0056] calculate and The Pearson correlation coefficient between them. It satisfies the following formula: .

[0057] in, These are the means of the data within the window. It should be noted that the above Pearson correlation coefficient... The calculation formula applies when the denominator is greater than a preset stability threshold (e.g., 10). −6 In the case where the denominator is less than the preset numerical stability threshold, it is determined that there is no effective change in power during the current time period, and the power control trajectory angle of the previous moment is used.

[0058] Set linearity threshold (For example, if 0.90 is chosen, a value of 0.85-0.95 is recommended; use a higher value when the control strategy is stable and a lower value when there are rapid changes.) If This indicates that the changes in active and reactive power during this period show a significant linear relationship, the control strategy is stable, and the power control trajectory angle can be updated.

[0059] like This indicates that the data distribution is relatively discrete or there is no obvious linear control law. In this case, the angle value from the previous moment is directly used, that is, let = This effectively avoids angle calculation jitter during the transition process.

[0060] Step 302: When the correlation between the active power change data and the reactive power change data meets the preset linearity condition, determine the power control trajectory angle based on the data within the historical time window.

[0061] As one possible implementation, the specific calculation process of the power control trajectory angle is as follows: The active power change data within the historical time window are summed to obtain the active power sum; the reactive power change data within the historical time window are summed to obtain the reactive power sum; the active power sum is used as the component of the active power coordinate axis direction, and the reactive power sum is used as the component of the reactive power coordinate axis direction, forming a composite vector; the angle between the composite vector and the positive direction of the active power coordinate axis is calculated using a preset angle determination algorithm, and this angle is used as the power control trajectory angle to characterize the dominant power change direction of the microgrid.

[0062] As an example, power control trajectory angle Satisfy the following formula: .

[0063] Where atan2(y,x) is a two-parameter arctangent function, and its return value range is... The summation operation is equivalent to performing vector synthesis on the vector formed by all scattered points within the window. The direction of the synthesized vector is the statistical direction of the principal components.

[0064] Step 303: When the correlation does not meet the preset linearity condition, the power control trajectory angle of the previous moment is taken as the power control trajectory angle of the current moment.

[0065] Based on the above technical solution, this embodiment ensures that the power control trajectory angle is updated only during periods when the control strategy is stable through a linearity pre-verification mechanism, thus avoiding angle calculation distortion caused by sudden changes in the control strategy or abnormal data fluctuations. By using weighted summation and a two-parameter arctangent function to calculate the power control trajectory angle, the robustness of the algorithm to noise is enhanced.

[0066] In one possible implementation, after determining the power control trajectory angle, the process of rotating the power change based on the power control trajectory angle to obtain the control axis power component and the deviation axis power component specifically includes: Based on the calculated trajectory angle Construct a local rotating coordinate system that follows the current movement. The physical power difference components at the current moment... and When projected onto this coordinate system, it is decomposed into two orthogonal eigencomponents.

[0067] Define the rotation transformation matrix : .

[0068] The original data is transformed using this matrix to obtain the control axis power components. Off-axis power components : .

[0069] Specifically, it can be elaborated as follows: .

[0070] .

[0071] Control axis components This represents the projected length of the power change vector along the control trajectory. Since the inverter mainly operates along this trajectory, this component contains most of the energy fluctuation information, has a high signal-to-noise ratio, and is the main excitation source for parameter identification.

[0072] Off-axis components This represents the deviation of the power change vector from the control trajectory. Under ideal control, this value is zero; in practice, it mainly includes measurement noise or minor control jitter.

[0073] In one possible implementation, before determining the active voltage sensitivity of each microgrid to each monitoring node, the method further includes: Step 401: Construct the input feature vector for the current moment based on the control axis power components and off-axis power components of all microgrids at the current moment.

[0074] In some embodiments, to prevent the continuous expansion of the covariance matrix of the recursive algorithm (Windup phenomenon) during "power quiescent" periods such as nighttime (when photovoltaics do not generate electricity), stable periods of cloud cover, or windless periods, which are caused by the input feature vector containing only a small amount of background noise, thus leading to random drift in the parameter estimates, this application first verifies the effectiveness of the input feature vector.

[0075] Optionally, the input feature vector is a multi-dimensional vector formed by arranging the control axis power components and off-axis power components of all microgrids in a preset order. This vector is the input data of the regularized recursive estimation algorithm, and its dimension is twice the number of microgrids, which can comprehensively characterize the power decoupling characteristics of all microgrids at the current moment.

[0076] For example, input feature vector It can be represented as: .

[0077] Step 402: Calculate the norm of the input feature vector and compare the norm with the preset excitation dead zone threshold.

[0078] As an example, the norm of the input feature vector Satisfy the following formula: .

[0079] Step 403: When the norm is less than the preset excitation dead zone threshold, it is determined that the current moment is in the identification silence period, and the active voltage sensitivity determined in the previous moment remains unchanged.

[0080] Step 404: When the norm is greater than or equal to the preset excitation dead zone threshold, it is determined that there is effective excitation at the current moment, and the active voltage sensitivity of each microgrid to each monitoring node is determined.

[0081] Specifically, setting an excitation dead zone threshold (It is recommended to set the value to the maximum of the inverter's rated capacity. For example, for a 1MW microgrid, the threshold can be set to the per-unit value corresponding to 1-5kW.) like If the system is currently in a silent period and lacks the conditions for identification, then the sensitivity parameter at the current moment is directly set to the value at the previous moment, i.e., , while keeping the covariance matrix unchanged.

[0082] like The process involves determining the existence of a valid incentive and executing the procedure to determine the active voltage sensitivity of each microgrid to each monitoring node.

[0083] Optionally, the process of determining the active voltage sensitivity of each microgrid to each monitoring node is as follows: using the differential voltage response as the identification target and the control axis power component and off-axis power component as the identification input, the axial sensitivity coefficient is iteratively updated using a recursive estimation algorithm with a forgetting factor and regularization coefficient; the updated axial sensitivity coefficient and power control trajectory angle are restored by inverse transformation, and the active direction sensitivity component is extracted by inverse rotation transformation to determine the active voltage sensitivity.

[0084] Specifically, for each monitoring node m, a state estimation model is maintained. A parameter vector is defined. Includes the axial sensitivity coefficient to be identified: .

[0085] Following this, the axial sensitivity coefficient is iteratively updated. The algorithm's iterative update steps are as follows: Calculate prior prediction error Predict the current voltage increment on the distribution network using the estimated value from the previous moment, and compare it with the measured value to determine the prior prediction error. Satisfy the following formula: .

[0086] Update covariance matrix (Including regularization): The covariance update formula of conventional Recursive Least Squares (RLS) diverges when the input is weak. This embodiment uses a regularized correction form: .

[0087] in, This is the forgetting factor, used to adjust the length of time the algorithm remembers historical data. A suggested value range is [0.95, 0.99]. A smaller value results in faster tracking speed but weaker noise resistance. These are Tikhonov regularization coefficients, used to prevent matrix singularities, and... Units should be consistent; recommended value range [ The existence of this item ensures that even if data is missing in a certain direction (such as off-axis), the matrix is ​​still valid. It remains positive definite and bounded. I is a 2N*2N identity matrix.

[0088] Calculate the regularized gain vector: .

[0089] Update parameter estimates: .

[0090] Through the above process, the algorithm can stably output the axial sensitivity coefficient. and Due to the existence of the regularization term, for insufficient incentives... The algorithm tends to suppress its estimates within a smaller numerical range rather than allowing them to diverge, in order to conform to the physical fact that the off-axis components are mainly noise.

[0091] Calculate the active power-voltage sensitivity of the i-th microgrid to the m-th monitoring node. : .

[0092] This coefficient This study integrates the direct impact of active power regulation in microgrids and the coupled impact of accompanying reactive power regulation. Specifically: The term reflects the contribution of the active component to the voltage along the control trajectory; The correction addresses minor effects caused by control deviations or noise. In this way, the present invention successfully and indirectly reconstructs, under conditions of data collinearity, the key physical parameters describing "how many volts the node voltage rises for every 1kW increase in active power output in a microgrid (accompanied by a proportional change in reactive power)".

[0093] Based on the above technical solution, this application quantifies the power excitation intensity at the current moment by constructing an input feature vector and calculating its norm, and determines the identification quiescent period and effective excitation period by combining a preset excitation dead zone threshold. During the identification quiescent period, the active voltage sensitivity of the previous moment remains unchanged, avoiding algorithm iteration divergence and parameter drift caused by the lack of effective excitation, thus ensuring the stability of the active voltage sensitivity. Parameter identification is only performed during the effective excitation period, ensuring that the identification process has sufficient power change data to support it, thereby improving the identification accuracy of the active voltage sensitivity. This step adds a pre-verification mechanism to the regularized recursive estimation, which can effectively avoid parameter errors caused by invalid identification, making the obtained active voltage sensitivity more consistent with the actual operating state of the distribution network, thereby improving the accuracy of voltage safety constraint boundaries and the rationality of power allocation optimization, and further enhancing the stability of the coordinated dispatch of the main distribution microgrid.

[0094] In one possible implementation, the specific implementation process of constructing the voltage safety constraint boundary based on active voltage sensitivity, the current voltage of each monitoring node, and a preset voltage safety threshold includes: For any monitoring node, based on the current voltage of the monitoring node, the preset voltage safety threshold, the preset voltage safety margin, and the active voltage sensitivity of each microgrid to the monitoring node, a linear inequality constraint on the power setpoint of each microgrid is constructed; the combination of the linear inequality constraints corresponding to all monitoring nodes is used as the voltage safety constraint boundary.

[0095] Specifically, for each monitoring node m, the following linear inequality constraint is constructed: .in, The current measured voltage at node m; This represents the current measured active power of microgrid i. The active power setpoint for the next time step to be solved; The upper limit of voltage specified by the power grid (e.g., 231V or 1.05pu). This is a reserved voltage buffer. A value of 0.01-0.02 pu (i.e., 1% to 2% of the nominal voltage) is recommended. This margin ensures that even with some prediction error, the scheduling results will likely fall within the absolutely safe range.

[0096] Rearranging the above inequalities, we obtain the standard form for the decision variables: Among them, the constant term on the right-hand side .

[0097] For all M critical nodes in the distribution network, this set of inequalities together form a multidimensional convex polyhedron, namely the voltage-constrained power boundary. .

[0098] Based on the above technical solutions, this application introduces a voltage safety margin for each monitoring node and constructs linear inequality constraints on the microgrid power setpoint. This effectively offsets the identification error of active voltage sensitivity and the nonlinear error of grid operation, avoiding actual voltage exceeding limits due to errors and improving the reliability of voltage safety constraints. By combining the linear inequality constraints of all monitoring nodes into a voltage safety constraint boundary, comprehensive control over the voltage of all key nodes in the distribution network is achieved, avoiding the problem of local node voltage exceeding limits and ensuring the overall safety of distribution network voltage operation. At the same time, the linear inequality constraints are simple in form and can be quickly integrated into the subsequent power allocation optimization model, reducing the solution complexity of the model, improving the computational efficiency of power allocation, and adapting to the real-time requirements of coordinated dispatch of main and distribution microgrids. The linear constraints constructed based on active voltage sensitivity can accurately reflect the correlation between microgrid power regulation and node voltage, making the voltage safety constraint boundary more consistent with the actual operating characteristics of the distribution network and further improving the rationality of power allocation optimization.

[0099] In one possible implementation, a preset power allocation optimization model is solved to obtain the active power setpoint for each microgrid. This includes: establishing a power allocation optimization model with the objective functions of minimizing the power regulation amplitude of each microgrid and minimizing the deviation between the power of each microgrid and its preset economic optimal operating point; using the total power dispatch command as an equality constraint, and the voltage safety constraint boundary, the physical output limit constraint of each microgrid, and the trust domain step size constraint used to limit the single power regulation amplitude as inequality constraints, which together constitute the solution conditions for the power allocation optimization model; and solving the power allocation optimization model that satisfies the solution conditions to obtain the active power setpoint for each microgrid.

[0100] Specifically, after receiving the total active power dispatch instruction for the distribution network issued by the upper-level main grid for the next time step, a quadratic programming (QP) model is established. Under the premise of satisfying the total power dispatch instruction, the optimal and safe microgrid power combination is obtained. The specific model construction and solution process is as follows: Using the active power setpoint of each microgrid as the optimization variable, the optimization variable vector is constructed as follows: Where N is the number of microgrids in the distribution network. Let be the active power setpoint for the i-th microgrid. A weighted least squares approach is used to construct the objective function, which balances the stability of microgrid power regulation with the economic efficiency of distribution network operation. The objective function is: The first item The aim is to minimize power regulation amplitude and avoid frequent large-amplitude operations by the equipment; the second item The aim is to guide microgrids closer to their economically optimal operating point. (Such as Maximum Power Point Tracking (MPPT) points or economic dispatch points); These are weighting coefficients, which can be set according to actual needs (e.g., taking...). ).

[0101] To ensure the safety, compliance, and effectiveness of power allocation results, multi-dimensional constraints are set for the quadratic programming model, specifically including: total power balance constraints (response to the main network). Voltage safety boundary constraints: Equipment physical constraints: .in These represent the lower and upper limits of the inverter's physical output. Trust domain step size constraint: due to sensitivity parameters... It is locally linear. In order to ensure the effectiveness of the linear prediction model, the power change amplitude of a single scheduling must be limited (i.e., limited to the trust region). The trust domain step size This is the maximum permissible single-step adjustment (recommended to be 10% to 20% of the inverter's rated capacity). This constraint prevents the optimizer from searching for solutions far from the current operating point, thereby significantly improving the actual execution accuracy of the scheduling results. Trust domain step size. Based on the linear range of the sensitivity parameters and the inverter response speed, it is recommended to initially set the power setpoint to 10% of the rated capacity, adjusting it according to actual operating results. The optimal power setpoint for each microgrid can be obtained by solving the above model using a standard quadratic programming solver (such as Operator Splitting Quadratic Program (OSQP), IBM ILOG CPLEX Optimization Studio (CPLEX), Gurobi Optimizer (Gurobi), etc.). The above-mentioned optimization variables, objective function, and constraints together constitute a complete quadratic programming model. The model is then solved using a standard quadratic programming solver. The resulting optimization variable vector Pset is the optimal active power setpoint for each microgrid. This setpoint can accurately respond to the dispatch instructions from the upper-level main grid while taking into account the voltage safety of the distribution network, the safety of equipment operation, the stability of regulation, and the economic efficiency of operation.

[0102] Based on the above technical solutions, this application takes minimizing the power regulation amplitude and minimizing the deviation from the economically optimal operating point as objective functions. While achieving voltage safety and responsiveness to superior dispatch instructions, it also considers the operational stability of microgrid equipment and the overall operational economy of the distribution network, realizing multi-objective collaborative optimization. By setting equality constraints for total power dispatch instructions, inequality constraints for voltage safety constraint boundaries, physical output limit constraints, and trust domain step size constraints, comprehensive and rigorous solution conditions are constructed. This ensures both the accurate overall response of the distribution network to superior dispatch instructions and the effectiveness of node voltage safety, equipment operation safety, and voltage constraints, avoiding the problem of unreasonable dispatch caused by single constraints. A quadratic programming model is used for solution, which has a simple model structure, high solution efficiency, and can quickly obtain the active power setpoint of each microgrid, adapting to the real-time requirements of main-distribution microgrid collaborative dispatch. At the same time, the standardized model construction and solution method can adapt to distribution networks of different scales and structures, improving the versatility and practicality of the method, and enabling main-distribution microgrid collaborative dispatch to further achieve economic and stable operation on the basis of safety and accuracy.

[0103] Please see Figure 2 This document illustrates a system architecture diagram of a primary-distribution microgrid collaborative optimization scheduling system according to an embodiment of this application. The system includes: a data acquisition and synchronization unit 201, a power decoupling processing unit 202, a sensitivity identification unit 203, a constraint boundary construction unit 204, and a power optimization allocation unit 205. The units communicate bidirectionally via communication links to ensure real-time interaction of acquired data and analysis results. These communication links can employ wired or wireless transmission methods to meet the communication needs of different monitoring scenarios.

[0104] The data acquisition and synchronization unit 201 is used to acquire the gate voltage of the upstream power grid connected to the distribution network, the voltage of each monitoring node in the distribution network, and the active and reactive power of each microgrid connection point in the distribution network, and to complete the time synchronization of the data.

[0105] The power decoupling processing unit 202 is used to determine the power control trajectory angle of each microgrid based on the power changes of active power and reactive power, and to perform coordinate rotation on the power changes based on the power control trajectory angle to obtain the control axis power component and the deviation axis power component.

[0106] The sensitivity identification unit 203 is used to determine the active voltage sensitivity of each microgrid to each monitoring node based on the control axis power component, the off-axis power component, and the differential voltage response of each monitoring node voltage relative to the threshold voltage.

[0107] The constraint boundary construction unit 204 is used to construct voltage safety constraint boundaries based on active voltage sensitivity, the current voltage of each monitoring node, and a preset voltage safety threshold.

[0108] The power optimization allocation unit 205 is used to respond to the total power dispatch command received from the upper-level power grid, and solve the preset power allocation optimization model with the total power dispatch command and voltage safety constraint boundary as constraints to obtain the active power setpoint of each microgrid.

[0109] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0110] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for coordinated optimization scheduling of primary and distribution microgrids, characterized in that, The method includes: Obtain the gate voltage of the upstream power grid connected to the distribution network, the voltage of each monitoring node in the distribution network, and the active and reactive power of each microgrid connection point in the distribution network; Based on the power changes in active and reactive power, the power control trajectory angle of each microgrid is determined, and the power changes are rotated in coordinates based on the power control trajectory angle to obtain the control axis power component and the deviation axis power component. Based on the control axis power component, the off-axis power component, and the differential voltage response of each monitoring node voltage relative to the cut-off voltage, the active voltage sensitivity of each microgrid to each monitoring node is determined. Based on the active voltage sensitivity, the current voltage of each monitoring node, and the preset voltage safety threshold, a voltage safety constraint boundary is constructed. In response to receiving the total power dispatch command from the upper-level power grid, the preset power allocation optimization model is solved using the total power dispatch command and the voltage security constraint boundary as constraints to obtain the active power setpoint for each microgrid.

2. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 1, characterized in that, Based on the changes in active and reactive power, the power control trajectory angle of each microgrid is determined, including: Linearity pre-verification is performed on the active power change data and reactive power change data of a single microgrid within a preset historical time window; When the correlation between the active power change data and the reactive power change data meets the preset linearity condition, the power control trajectory angle is determined based on the data within the historical time window. When the correlation does not meet the preset linearity condition, the power control trajectory angle of the previous moment is used as the power control trajectory angle of the current moment.

3. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 2, characterized in that, Determining the power control trajectory angle based on data within the historical time window includes: The active power change data within the historical time window are summed to obtain the active power cumulative sum; The reactive power change data within the historical time window are summed to obtain the reactive power sum. The sum of active power as the component of the active power coordinate axis and the sum of reactive power as the component of the reactive power coordinate axis constitute a composite vector. The angle between the synthesized vector and the positive direction of the active power coordinate axis is calculated by a preset angle determination algorithm, and the angle is used as the power control trajectory angle to characterize the power-dominant change direction of the microgrid.

4. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 1, characterized in that, Before determining the active voltage sensitivity of each microgrid to each monitoring node, the method further includes: Based on the control axis power components and off-axis power components of all microgrids at the current moment, construct the input feature vector at the current moment; Calculate the norm of the input feature vector and compare the norm with a preset excitation dead zone threshold; When the norm is less than the preset excitation dead zone threshold, it is determined that the current moment is in the identification silence period, and the active voltage sensitivity determined in the previous moment remains unchanged. When the norm is greater than or equal to the preset excitation dead zone threshold, it is determined that there is a valid excitation at the current moment, and the active voltage sensitivity of each microgrid to each monitoring node is determined.

5. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 4, characterized in that, Determine the active voltage sensitivity of each microgrid to each monitoring node, including: Using the differential voltage response as the identification target and the control axis power component and the off-axis power component as the identification input, the axial sensitivity coefficient is iteratively updated using a recursive estimation algorithm with a forgetting factor and regularization coefficient. The updated axial sensitivity coefficient and the power control trajectory angle are inversely transformed and restored to determine the active voltage sensitivity.

6. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 1, characterized in that, Based on the active voltage sensitivity, the current voltage of each monitoring node, and the preset voltage safety threshold, a voltage safety constraint boundary is constructed, including: For any monitoring node, a linear inequality constraint on the power setpoint of each microgrid is constructed based on the current voltage of the monitoring node, the preset voltage safety threshold, the preset voltage safety margin, and the active voltage sensitivity of each microgrid to the monitoring node. The combination of the linear inequality constraints corresponding to all monitoring nodes is used as the voltage safety constraint boundary.

7. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 1, characterized in that, Solving the preset power allocation optimization model yields the active power setpoints for each microgrid, including: A power allocation optimization model is established with the objective functions of minimizing the power regulation amplitude of each microgrid and minimizing the deviation between the power of each microgrid and its preset economic optimal operating point. The total power dispatch command is used as an equality constraint, and the voltage safety constraint boundary, the physical output limit constraint of each microgrid, and the trust domain step size constraint used to limit the magnitude of a single power adjustment are used as inequality constraints, which together constitute the solution conditions of the power allocation optimization model. The power allocation optimization model that satisfies the solution conditions is solved to obtain the active power setpoint for each microgrid.

8. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 1, characterized in that, The process for determining the power change includes: Calculate the timing difference components of the voltage at each monitoring node, the timing difference components of the threshold voltage, and the timing difference components of the active power and reactive power, respectively. The time-series difference component between the active power and reactive power is used as the power change.

9. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 8, characterized in that, The process of determining the differential voltage response includes: For any monitoring node, the time-series difference component of the voltage at the monitoring node is calculated to be different from the time-series difference component of the threshold voltage. The result of the calculation is used as the differential voltage response corresponding to the monitoring node. The differential voltage response is used to characterize the relative voltage change caused by power regulation within the distribution network.

10. The method for coordinated optimization scheduling of primary and distribution microgrids according to claim 1, characterized in that, Acquire the gate voltage of the upstream power grid connected to the distribution network, the voltage of each monitoring node within the distribution network, and the active and reactive power of each microgrid connection point within the distribution network, including: Using a preset scheduling cycle as the time reference, the current voltage amplitude at the control point, the voltage amplitude at each monitoring node, and the voltage amplitude, active power value, and reactive power value at each microgrid grid connection point are collected. When the timestamp of any collected data deviates from the current time, linear interpolation is used to synchronize all collected data to the same time segment.

Images