On-line calculation method and system for grid-connected point voltage sensitivity under reverse power working condition

Abstract

The application provides a grid-connected point voltage sensitivity online calculation method and system under reverse power conditions, the method comprises the following steps: S1, obtaining parameters of a target power distribution network at a grid-connected point and preprocessing to obtain grid-connected point time series data; S2, judging the reverse power conditions based on the grid-connected point time series data, and constructing a disturbance sample sequence in the time period of the determined reverse power conditions; S3, establishing and training a grid-connected point voltage sensitivity online calculation model; S4, inputting real-time samples into the trained voltage sensitivity online calculation model, outputting the sensitivity of the grid-connected point voltage to active and reactive power injection, and generating corresponding sensitivity curves and effectiveness indexes. The application can realize fast online calculation of the grid-connected point voltage sensitivity under reverse power conditions, improve the stability and reliability of the sensitivity results, provide a basis for power distribution network voltage out-of-limit early warning, reactive power optimization and active power limiting, and ensure the safe and reliable operation of the power distribution network.

Description

Technical Field

[0001] This invention relates to the field of power system and distribution network operation analysis technology, specifically to an online calculation method and system for grid connection point voltage sensitivity under reverse power conditions. Background Technology

[0002] As the penetration rate of distributed power sources such as photovoltaic power generation in distribution networks continues to increase, the power flow in distribution networks is gradually shifting from traditional unidirectional power supply to bidirectional flow, leading to more frequent reverse power conditions. Under reverse power conditions, the voltage at the grid connection point and its adjacent nodes is more sensitive to changes in active / reactive power injection, which can easily cause problems such as voltage exceeding limits, reactive power regulation conflicts, and malfunctions in protection and control systems. Therefore, it is necessary to monitor the sensitivity of the grid connection point voltage to power disturbances in real time to support voltage control, reactive power optimization, flexible load / energy storage coordination, and the formulation of power limiting strategies.

[0003] Current voltage sensitivity calculations largely rely on offline power flow models or analytical derivations based on the Jacobian matrix, typically requiring complete network parameters, operational cross-sections, and significant computational costs. When distribution network topology changes frequently, equipment parameters are uncertain, measurements are missing, or noise levels are high, sensitivity results are prone to deviation and struggle to meet the real-time and robustness requirements of online applications. On the other hand, some methods employ multiple power flow disturbances or simulation scans to obtain sensitivity, which, while highly applicable, incurs additional computational overhead in online scenarios and may affect the timeliness of scheduling decisions.

[0004] Therefore, how to combine real-time measurement and operational section characteristics under reverse power conditions to achieve rapid, online, and stable calculation of grid connection point voltage sensitivity, and improve the ability to predict voltage risks and the effectiveness of control strategies, has become an urgent technical problem to be solved. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide an online calculation method and system for grid connection point voltage sensitivity under reverse power conditions.

[0006] A method for online calculation of grid connection point voltage sensitivity under reverse power conditions, provided by the present invention, includes:

[0007] Step S1: Obtain the parameters of the target distribution network at the grid connection point and perform preprocessing to obtain the grid connection point time series data;

[0008] Step S2: Based on the grid connection point time series data, identify the reverse power condition and construct a disturbance sample sequence within the time period in which the reverse power condition is determined;

[0009] Step S3: Establish and train an online calculation model for grid point voltage sensitivity based on the perturbation sample sequence;

[0010] Step S4: Input the real-time samples into the trained online voltage sensitivity calculation model, output the sensitivity of the grid connection point voltage to active and reactive power injection, and generate the corresponding sensitivity curve and effectiveness index.

[0011] Preferably, the parameters at the grid connection point include real-time measurement data and network basic parameters; the grid connection point time series data includes grid connection point voltage time series data and grid connection point power injection time series data; the preprocessing includes time alignment and anomaly cleaning of the real-time measurement data.

[0012] Preferably, step S1 includes:

[0013] Obtain the voltage phasor at the grid connection point With current phasor and equivalent impedance of lines / transformers And calculate the power injection sequence at the grid connection point based on the complex power relationship:

[0014]

[0015] in, For sampling time index; For conjugate; Power injection for grid connection point; and These are active power injection and reactive power injection, respectively. The imaginary unit; and These are the equivalent resistance and reactance, respectively; and for , , Perform outlier removal and missing value completion to obtain cleaned time series data.

[0016] Preferably, step S2 includes:

[0017] Based on the power direction of the power injection time series data at the grid connection point, the inverse power condition is determined, and a perturbation sample sequence for online sensitivity estimation is constructed within the time period in which the inverse power condition is determined.

[0018] Active power injection at grid connection point The direction is used as the criterion, and the following formula is used to determine the reverse power condition:

[0019]

[0020] in, This is a marker for reverse power operation. The inverse power discrimination threshold, It is an indicator function;

[0021] Within the inverse power condition sample, a perturbation sample sequence is constructed using differential methods:

[0022]

[0023]

[0024]

[0025] in, For grid connection point voltage disturbance, and The disturbances are injected into the active and reactive power respectively; within the rolling window, sample points that meet the disturbance amplitude threshold are selected as valid disturbance samples.

[0026] Preferably, step S3 includes the following sub-steps:

[0027] Step S3.1: Based on the disturbance sample sequence, establish an online calculation model for grid connection point voltage sensitivity under inverse power conditions, and use historical samples or rolling window samples to perform initial training or initial fitting of the model to obtain an initial sensitivity estimation model.

[0028] Step S3.2: Introduce a consistency verification mechanism based on power flow / state estimation or metering power direction, construct a loss function and update the parameters of the initial sensitivity estimation model online to obtain the trained online voltage sensitivity calculation model.

[0029] Preferably, step S3.1 includes:

[0030] Establish a local linear sensitivity model:

[0031]

[0032] in, This is the sensitivity coefficient of the grid connection point voltage to active power injection. This is the sensitivity coefficient of the grid connection point voltage to reactive power injection. For residual terms;

[0033] This model is used to characterize the local linear coupling relationship between voltage and power under inverse power conditions;

[0034] Scroll window The effective perturbation samples within the range form the regression matrix. With output vector The initial sensitivity estimate is obtained using weighted least squares:

[0035]

[0036] in, This is the initial estimated parameter vector; This is the weight matrix. For sample weights; This is a transpose.

[0037] Preferably, step S3.2 includes:

[0038] For the current moment or scrolling window Internal samples are obtained by using power flow / state estimation results or recordings from metering devices to obtain the reference change in grid-connected point voltage. and power direction verification quantity ;

[0039] The online model's prediction of voltage disturbances is expressed as: And construct the loss function:

[0040]

[0041] in, The objective function for online optimization; These are sample weights, used to enhance the constraints on recent or perturbed samples; These are consistency constraint coefficients used to balance the voltage fitting term and the power direction consistency term; It is an indicator function; This is the inverse power discrimination threshold.

[0042] Preferably, step S4 includes:

[0043] When real-time samples are input into the trained online voltage sensitivity calculation model, the current time... and its short-term scrolling window The measurement data from the grid connection points within the system are continuously aligned for execution time and anomaly cleaned, and power disturbance samples are constructed accordingly. , and voltage disturbances , Inputting it into the trained model yields the sensitivity output at the current time. and And record short-time sensitivity curves by rolling records according to time index. .

[0044] Preferably, step S4 further includes:

[0045] The effectiveness index is defined as follows:

[0046]

[0047] in, For the first Effectiveness indicators for time-sensitivity estimation; This represents the voltage disturbance within the window. , Inject disturbances into the active / reactive power within the window; , These are the online sensitivity outputs of the grid connection point voltage to active / reactive injection, respectively. For window The mean of the internal voltage disturbance.

[0048] An online calculation system for grid connection point voltage sensitivity under reverse power conditions, provided by the present invention, includes:

[0049] The data acquisition and preprocessing module is used to acquire real-time measurement data and network basic parameters of the target distribution network at the grid connection point and perform preprocessing to obtain grid connection point voltage time series data and grid connection point power injection time series data.

[0050] The inverse power condition discrimination and disturbance sample construction module determines the inverse power condition based on the power direction of the power injection time series data at the grid connection point and constructs a disturbance sample sequence.

[0051] The sensitivity initial fitting module establishes an online calculation model of grid connection point voltage sensitivity under inverse power conditions based on the disturbance sample sequence and performs initial training on it.

[0052] The online verification and parameter update module is used to construct the loss function and update the parameters of the model online to obtain the trained online calculation model of grid connection point voltage sensitivity.

[0053] The online output and index generation module takes real-time samples as input to the trained online calculation model of grid connection point voltage sensitivity, and outputs sensitivity results, sensitivity curves, and effectiveness indices.

[0054] Preferably, the preprocessing includes time alignment and anomaly cleaning of the real-time measurement data.

[0055] Preferably, the data acquisition and preprocessing module includes:

[0056] Obtain the voltage phasor at the grid connection point With current phasor and equivalent impedance of lines / transformers And calculate the power injection sequence at the grid connection point based on the complex power relationship:

[0057]

[0058] in, For sampling time index; For conjugate; Power injection for grid connection point; and These are active power injection and reactive power injection, respectively. The imaginary unit; and These are the equivalent resistance and reactance, respectively; and for , , Perform outlier removal and missing value completion to obtain cleaned time series data.

[0059] Preferably, the inverse power condition discrimination and disturbance sample construction module includes:

[0060] Based on the power direction of the power injection time series data at the grid connection point, the inverse power condition is determined, and a perturbation sample sequence for online sensitivity estimation is constructed within the time period in which the inverse power condition is determined.

[0061] Active power injection at grid connection point The direction is used as the criterion, and the following formula is used to determine the reverse power condition:

[0062]

[0063] in, This is a marker for reverse power operation. The inverse power discrimination threshold, It is an indicator function;

[0064] Within the inverse power condition sample, a perturbation sample sequence is constructed using differential methods:

[0065]

[0066]

[0067]

[0068] in, For grid connection point voltage disturbance, and The disturbances are injected into the active and reactive power respectively; within the rolling window, sample points that meet the disturbance amplitude threshold are selected as valid disturbance samples.

[0069] Preferably, the initial sensitivity fitting module includes:

[0070] Establish a local linear sensitivity model:

[0071]

[0072] in, This is the sensitivity coefficient of the grid connection point voltage to active power injection. This is the sensitivity coefficient of the grid connection point voltage to reactive power injection. For residual terms;

[0073] This model is used to characterize the local linear coupling relationship between voltage and power under inverse power conditions;

[0074] Scroll window The effective perturbation samples within the range form the regression matrix. With output vector The initial sensitivity estimate is obtained using weighted least squares:

[0075]

[0076] in, This is the initial estimated parameter vector; This is the weight matrix. For sample weights; This is a transpose.

[0077] Preferably, the online verification and parameter update module includes:

[0078] For the current moment or scrolling window Internal samples are obtained by using power flow / state estimation results or recordings from metering devices to obtain the reference change in grid-connected point voltage. and power direction verification quantity ;

[0079] The online model's prediction of voltage disturbances is expressed as: And construct the loss function:

[0080]

[0081] in, The objective function for online optimization; These are sample weights, used to enhance the constraints on recent or perturbed samples; These are consistency constraint coefficients used to balance the voltage fitting term and the power direction consistency term; It is an indicator function; This is the inverse power discrimination threshold.

[0082] Preferably, the online output and indicator generation module includes:

[0083] When real-time samples are input into the trained online voltage sensitivity calculation model, the current time... and its short-term scrolling window The measurement data from the grid connection points within the system are continuously aligned for execution time and anomaly cleaned, and power disturbance samples are constructed accordingly. , and voltage disturbances , Inputting it into the trained model yields the sensitivity output at the current time. and And record short-time sensitivity curves by rolling records according to time index. .

[0084] Preferably, the online output and indicator generation module further includes:

[0085] The effectiveness index is defined as follows:

[0086]

[0087] in, For the first Effectiveness indicators for time-sensitivity estimation; This represents the voltage disturbance within the window. , Inject disturbances into the active / reactive power within the window; , These are the online sensitivity outputs of the grid connection point voltage to active / reactive injection, respectively. For window The mean of the internal voltage disturbance.

[0088] Compared with the prior art, the present invention has the following beneficial effects:

[0089] 1. The method provided by this invention is driven by real-time measurement. Based on the identification of inverse power conditions, it constructs voltage-power disturbance samples and performs rolling estimation. Through consistency verification and online parameter updates, it improves the stability and reliability of sensitivity results. It can realize rapid online calculation of grid connection point voltage sensitivity under topology changes, measurement noise and operating condition fluctuations.

[0090] 2. This invention enables rapid online calculation of grid connection point voltage sensitivity under reverse power conditions, improving the stability and reliability of sensitivity results. It provides a basis for control decisions such as voltage over-limit early warning, reactive power optimization, and active power limiting in distribution networks, ensuring the safe and reliable operation of distribution networks.

[0091] 3. In practical applications, this invention can be used for voltage over-limit risk prediction, reactive power compensation and active power limiting strategy formulation, and energy storage / flexible load regulation, thereby improving the timeliness and effectiveness of regulation decisions. Attached Figure Description

[0092] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0093] Figure 1 This is a flowchart of the calculation method of the present invention.

[0094] Figure 2 This is a structural diagram of the voltage sensitivity estimation model in this invention.

[0095] Figure 3 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0096] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0097] A method for online calculation of grid connection point voltage sensitivity under reverse power conditions, comprising:

[0098] Step S1: Obtain the parameters of the target distribution network at the grid connection point and perform preprocessing to obtain the grid connection point time series data;

[0099] Step S2: Based on the grid connection point time series data, identify the reverse power condition and construct a disturbance sample sequence within the time period in which the reverse power condition is determined;

[0100] Step S3: Establish and train an online calculation model for grid point voltage sensitivity based on the perturbation sample sequence;

[0101] Step S4: Input the real-time samples into the trained online voltage sensitivity calculation model, output the sensitivity of the grid connection point voltage to active and reactive power injection, and generate the corresponding sensitivity curve and effectiveness index.

[0102] In one embodiment, the specific steps include:

[0103] Step S1: Obtain real-time measurement data and network basic parameters of the target distribution network at the grid connection point and perform preprocessing to obtain grid connection point voltage time series data and grid connection point power injection time series data;

[0104] Step S2: Based on the power direction of the power injection time series data at the grid connection point, determine the inverse power condition, and construct a perturbation sample sequence for online sensitivity estimation within the time period in which the inverse power condition is determined;

[0105] Step S3: Based on the disturbance sample sequence, establish an online calculation model for grid connection point voltage sensitivity under inverse power conditions, and use historical samples or rolling window samples to perform initial training or initial fitting of the model to obtain an initial sensitivity estimation model.

[0106] Step S4: Introduce a consistency verification mechanism based on power flow / state estimation or metering power direction, construct a loss function and update the parameters of the initial sensitivity estimation model online to obtain the trained online voltage sensitivity calculation model.

[0107] Step S5: Input the real-time samples into the trained online voltage sensitivity calculation model, output the sensitivity results of the grid connection point voltage to active / reactive power injection, and generate the sensitivity curve and effectiveness index at the current time and within the short-term rolling interval.

[0108] Example 1

[0109] like Figure 1 As shown, an online calculation method for grid connection point voltage sensitivity under reverse power conditions includes:

[0110] Step 1: Obtain real-time measurement data and network basic parameters of the target distribution network at the grid connection point, and perform time alignment and anomaly cleaning on the measurement data to obtain the grid connection point voltage time series data and grid connection point power injection time series data.

[0111] In this embodiment, the method for obtaining real-time measurement data and network basic parameters of the target distribution network at the grid connection point is as follows: The real-time measurement data and network basic parameters at the target distribution network's grid connection point (PCC) include the effective values ​​or phasors of the three-phase voltages at the grid connection point. Effective value or phasor of grid connection point current Power factor / phase angle and switching timestamps, etc.; basic network parameters include the equivalent line impedance from the grid connection point to the upstream power grid. ( The imaginary unit; and These are equivalent resistance and reactance, respectively; and transformer turns ratio / tap. The measured data includes the rated capacity and base voltage of the line / transformer. Then, the measurement data is time-aligned to ensure uniform sampling granularity. Resample the original asynchronous measurement sequence at each target time. The aligned grid-connected point voltage and current sequences are obtained by using a sliding window averaging method. For example, voltage alignment can be written as:

[0112]

[0113] in, For the first The grid connection point voltage of each aligned sampling point (which can be phasor amplitude or phasor). This is the original voltage measurement sequence. To standardize the sampling time window length, For the first Each aligned sampling time; for current It was obtained in the same way. After time alignment is completed, the grid-connected point power injection timing data is calculated based on the electrical power relationship using the grid-connected point voltage and current, employing a complex power definition:

[0114]

[0115] in, For the grid connection point at the first Complex power injection at each sampling time, For active power injection, For reactive power injection, The imaginary unit, Indicates conjugate operation; and For the voltage and current phasors (or equivalent phasors) at the grid connection point, the sign convention for the power direction at the grid connection point is "sending to the upper-level grid" as positive injection and "receiving from the upper-level grid" as negative injection (which is usually manifested in reverse power conditions). Or according to the corresponding reverse symbol agreed upon on site). Finally, align the... and The sequence is cleaned to suppress the impact of measurement spikes, communication packet loss and abrupt changes after interpolation on subsequent online sensitivity calculations.

[0116] In this embodiment, a robust cleaning rule based on the median and median absolute difference (MAD) is adopted:

[0117]

[0118] in, This represents the sequence to be cleaned (which can be selected). , or ), This is the result after cleaning. For the sliding window mid-width operator, the window half-width is... (Window length) ), The absolute median difference This is the threshold coefficient (used to control the anomaly detection strength). Through the above time alignment, power injection calculation, and anomaly cleaning, the grid-connected point voltage time-series data for subsequent online sensitivity estimation can be obtained. and grid connection point power injection timing data , .

[0119] Step 2: Based on the power direction at the grid connection point, identify the inverse power condition and construct a disturbance sample sequence for online sensitivity estimation, and extract the feature quantities characterizing the voltage-power coupling relationship;

[0120] First, the power direction is determined based on the power injection time series data at the grid connection point, using the active power injection at the grid connection point as an example. Using the sign and threshold as inverse power condition identification conditions, an inverse power discriminant is constructed:

[0121]

[0122] in, For the aligned discrete time index; For the first The active power injected from the grid connection point to the upper-level power grid at any given time (the notation convention is that "sending to the upper-level power grid" is positive); This is the inverse power discrimination threshold (used to suppress false judgments caused by noise near the zero point); This is an indicator function; it takes the value 1 when the condition inside the parentheses is true, and 0 otherwise. At that time, it was considered that the grid connection point was in reverse power operation condition (or reverse power risk operation condition). The system is considered to be in a non-reverse power condition. To further improve the electrical consistency of the operating condition determination, cross-verification can be performed using the power direction calculated from the voltage and current phasors, i.e., utilizing... Received The power direction is consistent with that recorded by the metering / protection device to avoid misjudgments caused by missing measurements or inconsistent symbol conventions. For sampling time / discrete time index; For the grid connection point PCC at time Voltage phasors; For the grid connection point PCC at time Current phasor; This is a conjugate operation; For grid connection points at time Complex power injection. To take the real part of the complex number; For grid connection points at time Active power injection.

[0123] Then, a perturbation sample sequence for online sensitivity estimation is constructed within the time period defined by the inverse power condition. Considering that voltage sensitivity essentially describes the "instantaneous response of voltage to power perturbations," therefore... , , Samples are constructed using differential perturbation, with perturbation vectors formed by adjacent sampling points:

[0124]

[0125]

[0126]

[0127] in, This is the disturbance to the grid connection point voltage (which can be either voltage amplitude disturbance or phasor amplitude disturbance). , These are the active and reactive power injection disturbances at the grid connection point, respectively; only when and or Only when the disturbance threshold is exceeded is the specified moment included in the valid disturbance sample to avoid small measurement noise dominating sensitivity estimation. To suppress occasional spikes, a rolling window can be used. The internal collection of perturbation samples forms the perturbation sample set required for online estimation. ,in This is the window length.

[0128] Feature quantities characterizing the "voltage-power coupling relationship" are extracted from disturbance samples to ensure that the subsequent online sensitivity calculation model can stably fit the voltage response under inverse power conditions. Since the distribution network is approximately linear over a short timescale, voltage and power disturbances can be characterized by linear sensitivity locally. Therefore, within a window... The local sensitivity characteristics of voltage to active and reactive power are obtained through least squares estimation:

[0129]

[0130] in, Indicates the first The voltage-active power sensitivity estimated within the rolling window at any time (reflecting the strength of the influence of changes in active power injection on voltage changes). Indicates voltage-reactive power sensitivity; The parameter to be estimated; , , For the first in the window One valid disturbance sample. To illustrate the electrical mechanism under inverse power conditions, basic network parameters (such as the equivalent impedance from the grid connection point to the upstream grid) can also be included. ) as auxiliary features and with Together they form a coupled feature vector, and The common input to the subsequent online sensitivity update stage ensures that the model maintains its sensitivity to the "resistance / reactance-dominated effect" even when the topology or equivalent impedance changes. Through the above-mentioned inverse power condition discrimination, disturbance sample construction, and coupled feature extraction, a high-quality sample sequence and feature quantity for online sensitivity estimation can be obtained, laying the foundation for the next step of online sensitivity calculation and parameter update.

[0131] Step 3: Establish an online calculation model for the voltage sensitivity at the grid connection point under reverse power conditions, and use historical samples or rolling window samples to perform initial training / fitting of the model to obtain the initial sensitivity estimation model.

[0132] like Figure 2 As shown, firstly, within the sample set identified as reverse power operating conditions, a local linear sensitivity model is constructed with grid connection point voltage disturbance as output and grid connection point active / reactive power injection disturbance as input. The response relationship of grid connection point voltage to power injection is expressed as follows:

[0133]

[0134] in, For the first The disturbance amount of the voltage amplitude (or voltage phasor amplitude) at the grid connection point at each sampling time; and These are the disturbances injected into the active and reactive power at the grid connection point, respectively. and These are the sensitivity coefficients of the grid connection point voltage to active power injection and reactive power injection (i.e., the target parameters to be calculated online). This includes residual terms caused by measurement noise, unmodeled dynamics, and parameter uncertainties. Then, the model is initially fitted using either a historical sample set or a rolling window sample set: when using a rolling window... When performing online estimation, the sample points within the window that satisfy the inverse power discrimination condition are used to form regression data, and the initial sensitivity estimate is obtained by weighted least squares:

[0135]

[0136] in, For each row of the regression input matrix, the corresponding window contains the first... Power perturbation of each sample; This is the output vector; The sample weight diagonal matrix, This is used to reflect the online estimation needs that "the closer to the current time, the greater the weight" or "the greater the disturbance amplitude, the greater the weight"; This represents the voltage sensitivity estimate obtained from the initial fitting. Through the above modeling and initial fitting, the initial sensitivity estimation model can be obtained, providing initial parameters and benchmark results for subsequent consistency verification, loss function, and online parameter optimization.

[0137] Step 4: Introduce a consistency verification mechanism based on power flow / state estimation or metering power direction, construct a loss function and update the model parameters online to obtain a trained online voltage sensitivity calculation model.

[0138] After obtaining the initial sensitivity estimation model, a consistency verification mechanism based on power flow / state estimation or metering power direction is introduced to constrain and correct the sensitivity output online. Specifically, this is applied to the current time or rolling window. Internal samples are obtained by using power flow / state estimation results or recordings from metering devices to obtain the reference change in grid-connected point voltage. and power direction verification quantity ,in This represents the grid connection point voltage disturbance calculated from power flow / state estimation or corrected by reliable measurements. This indicates the inverse power condition flag obtained from metered power direction, protection records, or master station criteria. Then, the online model's prediction of the voltage disturbance is expressed as... And construct a loss function that integrates "voltage fitting error + operating condition consistency constraint":

[0139]

[0140] in, The objective function for online optimization; These are sample weights, used to enhance the constraints on recent samples or samples with large disturbances; These are consistency constraint coefficients used to balance the voltage fitting term and the power direction consistency term;

[0141] It is an indicator function; This is the inverse power threshold. The parameters are then updated online based on this loss function, for example, using gradient descent or recursive least squares iterative updates.

[0142]

[0143] in, For the first The sensitivity parameter vector at the next update For online learning rate, This represents the gradient of the loss function with respect to the parameters. During the online update process, when the decrease in loss between two consecutive updates is less than a preset threshold or the change in parameters is less than a preset threshold, the model is considered to have converged, and the trained online voltage sensitivity calculation model is output. In subsequent runs, consistency checks and parameter fine-tuning are continuously performed using a rolling window approach to ensure that the sensitivity results remain stable and reliable under inverse power conditions and network parameter changes.

[0144] Step 5: Input the real-time samples into the trained model, output the sensitivity results of the grid connection point voltage to active / reactive power injection, and generate the sensitivity curves and effectiveness indicators for the current time and the short-term rolling interval.

[0145] When inputting real-time samples into the trained online voltage sensitivity calculation model, the first step is to calculate the current time... and its short-term scrolling window The measurement data from the grid connection points within the system are continuously aligned for execution time and anomaly cleaned, and power disturbance samples are constructed accordingly. , and voltage disturbances ( Then, it is input into the trained model to obtain the sensitivity output at the current time. and And record short-time sensitivity curves by rolling records according to time index. Meanwhile, to quantify the reliability of the sensitivity results under the current operating conditions, validity indicators such as the fitting residuals and coefficient of determination within the rolling window are calculated, and the validity indicators are defined as follows:

[0146]

[0147] in, For the first The effectiveness index of time sensitivity estimation (the closer the value is to 1, the better the fit and the more reliable the sensitivity). This represents the voltage disturbance within the window. , Inject disturbances into the active / reactive power within the window; , These are the online sensitivity outputs of the grid connection point voltage to active / reactive injection, respectively. For window The mean of the internal voltage disturbance.

[0148] The final output includes: current time sensitivity. Sensitivity curves within the short-term rolling interval, and corresponding effectiveness indicators. Used for constructing online monitoring and alarm criteria.

[0149] Step 6: Based on the calculation results, predict the risk of grid connection point voltage and regulate the controllable resources of the target distribution network. Based on the grid connection point voltage sensitivity calculation results output in Step S5, which include the current sensitivity, sensitivity change characteristics within the short-term rolling interval, and corresponding effectiveness indicators, construct a grid connection point voltage over-limit risk criterion. Calculate the predicted change or predicted value of the grid connection point voltage using the sensitivity results and real-time measurements and planned power changes. Compare the predicted grid connection point voltage with the minimum allowable value Umin and the maximum allowable value Umax of the preset voltage allowable range to determine the voltage risk level. When the predicted grid connection point voltage is close to the minimum allowable value Umin or the maximum allowable value Umax, and the distance threshold is less than the preset margin δ, output voltage warning information. When the predicted grid connection point voltage is less than the minimum allowable value Umin or greater than the maximum allowable value Umax, output voltage over-limit alarm information. The warning information and / or alarm information at least include the risk level, the expected over-limit direction, and the sensitivity result or sensitivity at the corresponding time. The system employs sensitivity curves and effectiveness indices, where effectiveness indices characterize the reliability of sensitivity results under current operating conditions to improve the reliability of alarm criteria. When a voltage over-limit risk is determined or an over-limit has already occurred, the system calculates the power regulation required to suppress the over-limit based on the sensitivity results and generates a control strategy. The control strategy preferably focuses on reactive power regulation, supplemented by active power limiting when necessary. The control strategy is then distributed to the controllable resources of the target distribution network for execution. The controllable resources include reactive power output of grid-connected inverters, adjustment of power factor setpoints or voltage control setpoints, switching or adjustment of reactive power compensation devices such as capacitors, reactors, SVG, or STATCOM, transformer tap adjustment, and active power limiting or ramp-up constraints for distributed power sources. After the control is executed, real-time measurement data such as grid-connected point voltage, current, and power injection are continuously collected, and the system returns to steps S1 to S6 for cyclical updates, achieving continuous online prediction and control of grid-connected point voltage risk under reverse power conditions.

[0150] Example 2

[0151] An online calculation system for grid connection point voltage sensitivity under reverse power conditions includes:

[0152] The data acquisition and preprocessing module is used to acquire real-time measurement data and network basic parameters of the target distribution network at the grid connection point and perform preprocessing to obtain grid connection point voltage time series data and grid connection point power injection time series data.

[0153] The inverse power condition discrimination and disturbance sample construction module determines the inverse power condition based on the power direction of the power injection time series data at the grid connection point and constructs a disturbance sample sequence.

[0154] The sensitivity initial fitting module establishes an online calculation model of grid connection point voltage sensitivity under inverse power conditions based on the disturbance sample sequence and performs initial training on it.

[0155] The online verification and parameter update module is used to construct the loss function and update the parameters of the model online to obtain the trained online calculation model of grid connection point voltage sensitivity.

[0156] The online output and index generation module takes real-time samples as input to the trained online calculation model of grid connection point voltage sensitivity, and outputs sensitivity results, sensitivity curves, and effectiveness indices.

[0157] like Figure 3 As shown, the data acquisition and preprocessing module is used to acquire real-time measurement data and network basic parameters at the target distribution network connection point, and to perform time alignment and anomaly cleaning on the measurement data to obtain the voltage time series data at the connection point. and grid connection point power injection timing data , The power injection at the grid connection point is calculated using complex power calculations. .

[0158] The inverse power condition discrimination and disturbance sample construction module is based on active power injection at the grid connection point. The power direction is determined when The system was determined to be inverse power condition, and the measurement sequence was differentially analyzed under inverse power condition to obtain perturbation samples. , , It is used to characterize voltage-power coupling relationships.

[0159] The initial sensitivity fitting module establishes a locally linear sensitivity model within the inverse power condition sample set.

[0160]

[0161] in This is the sensitivity coefficient of the grid connection point voltage to active power injection. This is the sensitivity coefficient of the grid connection point voltage to reactive power injection. For residual terms; and using historical samples or rolling window samples to... An initial fit is performed to obtain the initial sensitivity estimation model.

[0162] The validation and online parameter update module introduces a consistency validation mechanism and updates model parameters online.

[0163] Specifically, the model output is compared with the voltage reference perturbation obtained from the power flow / state estimation. Alternatively, perform consistency verification on the power direction records, construct a loss function, and adjust the parameter vector. Online updates, in the form of... The update stops when the loss converges or the difference in loss between adjacent rounds is less than the threshold, thus obtaining the trained online voltage sensitivity calculation model.

[0164] The online output and index generation module takes real-time samples as input to the trained model and outputs the sensitivity results at the current time. , A sensitivity curve is generated within a short-term rolling interval; simultaneously, a fitting validity index is calculated to evaluate the reliability of the results; the validity index can be expressed as the coefficient of determination. The fitting residuals within the rolling window are quantified and output together with the sensitivity curve for online monitoring and control decision-making.

[0165] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0166] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method for online calculation of grid connection point voltage sensitivity under reverse power conditions, characterized in that, include: Step S1: Obtain the parameters of the target distribution network at the grid connection point and perform preprocessing to obtain the grid connection point time series data; Step S2: Based on the grid connection point time series data, identify the reverse power condition and construct a disturbance sample sequence within the time period in which the reverse power condition is determined; Step S3: Establish and train an online calculation model for grid point voltage sensitivity based on the perturbation sample sequence; Step S4: Input the real-time samples into the trained online voltage sensitivity calculation model, output the sensitivity of the grid connection point voltage to active and reactive power injection, and generate the corresponding sensitivity curve and effectiveness index. The parameters at the grid connection point include real-time measurement data and basic network parameters; the grid connection point timing data includes grid connection point voltage timing data and grid connection point power injection timing data. The preprocessing includes time alignment and anomaly cleaning of the real-time measurement data; Step S2 includes: Based on the power direction of the power injection time series data at the grid connection point, the inverse power condition is determined, and a perturbation sample sequence for online sensitivity estimation is constructed within the time period in which the inverse power condition is determined. Active power injection at grid connection point The direction is used as the criterion, and the following formula is used to determine the reverse power condition: in, This is a marker for reverse power operation. The inverse power discrimination threshold, It is an indicator function; Within the inverse power condition sample, a perturbation sample sequence is constructed using differential methods: in, For grid connection point voltage disturbance, and The disturbances are injected into the active and reactive power respectively; within the rolling window, sample points that meet the disturbance amplitude threshold are selected as valid disturbance samples.

2. The online calculation method for grid connection point voltage sensitivity under reverse power conditions according to claim 1, characterized in that, Step S1 includes: Obtain the voltage phasor at the grid connection point With current phasor and equivalent impedance of lines / transformers And calculate the power injection sequence at the grid connection point based on the complex power relationship: in, For sampling time index; For conjugate; Power injection for grid connection point; and These are active power injection and reactive power injection, respectively. The imaginary unit; and These are the equivalent resistance and reactance, respectively; and for , , Perform outlier removal and missing value completion to obtain cleaned time series data.

3. The online calculation method for grid connection point voltage sensitivity under reverse power conditions according to claim 1, characterized in that, Step S3 includes the following sub-steps: Step S3.1: Based on the disturbance sample sequence, establish an online calculation model for grid connection point voltage sensitivity under inverse power conditions, and use historical samples or rolling window samples to perform initial training or initial fitting on the online calculation model for grid connection point voltage sensitivity to obtain an initial sensitivity estimation model. Step S3.2: Introduce a consistency verification mechanism based on power flow / state estimation or metering power direction, construct a loss function and update the parameters of the initial sensitivity estimation model online to obtain the trained online voltage sensitivity calculation model.

4. The online calculation method for grid connection point voltage sensitivity under reverse power conditions according to claim 3, characterized in that, Step S3.1 includes: Establish a local linear sensitivity model: in, This is the sensitivity coefficient of the grid connection point voltage to active power injection. This is the sensitivity coefficient of the grid connection point voltage to reactive power injection. For residual terms; The local linear sensitivity model is used to characterize the local linear coupling relationship between voltage and power under inverse power conditions; Scroll window The effective perturbation samples within the range form the regression matrix. With output vector The initial sensitivity estimate is obtained using weighted least squares: in, This is the initial estimated parameter vector; This is the weight matrix. For sample weights; This is a transpose.

5. The online calculation method for grid connection point voltage sensitivity under reverse power conditions according to claim 3, characterized in that, Step S3.2 includes: For the current moment or scrolling window Internal samples are obtained by using power flow / state estimation results or recordings from metering devices to obtain the reference change in grid-connected point voltage. and power direction verification quantity ; The online model's prediction of voltage disturbances is expressed as: And construct the loss function: in, The objective function for online optimization; These are sample weights, used to enhance the constraints on recent or perturbed samples; These are consistency constraint coefficients used to balance the voltage fitting term and the power direction consistency term; It is an indicator function; This is the inverse power discrimination threshold.

6. The online calculation method for grid connection point voltage sensitivity under reverse power conditions according to claim 3, characterized in that, Step S4 includes: When real-time samples are input into the trained online voltage sensitivity calculation model, the current time... and its short-term scrolling window The measurement data from the grid connection points within the system are continuously aligned for execution time and anomaly cleaned, and power disturbance samples are constructed accordingly. , and voltage disturbances , Inputting it into the trained model yields the sensitivity output at the current time. and And record short-time sensitivity curves by rolling records according to time index. .

7. The online calculation method for grid connection point voltage sensitivity under reverse power conditions according to claim 6, characterized in that, Step S4 further includes: The effectiveness index is defined as follows: in, For the first Effectiveness indicators for time-sensitivity estimation; This represents the voltage disturbance within the window. , Inject disturbances into the active / reactive power within the window; , These are the online sensitivity outputs of the grid connection point voltage to active / reactive injection, respectively. For window The mean of the internal voltage disturbance.

8. An online calculation system for grid connection point voltage sensitivity under reverse power conditions, characterized in that, include: The data acquisition and preprocessing module is used to acquire real-time measurement data and network basic parameters of the target distribution network at the grid connection point and perform preprocessing to obtain grid connection point voltage time series data and grid connection point power injection time series data. The inverse power condition discrimination and disturbance sample construction module determines the inverse power condition based on the power direction of the power injection time series data at the grid connection point and constructs a disturbance sample sequence. The sensitivity initial fitting module establishes an online calculation model of grid connection point voltage sensitivity under inverse power conditions based on the disturbance sample sequence and performs initial training on it. The online verification and parameter update module is used to construct the loss function and update the parameters of the model online to obtain the trained online calculation model of grid connection point voltage sensitivity. The online output and index generation module takes real-time samples as input to the trained online calculation model of grid connection point voltage sensitivity and outputs sensitivity results, sensitivity curves and effectiveness indices. The preprocessing of the data acquisition and preprocessing module includes time alignment and anomaly cleaning of real-time measurement data; The inverse power condition discrimination and disturbance sample construction module discriminates the inverse power condition based on the power direction of the power injection time series data at the grid connection point, and constructs a disturbance sample sequence for online sensitivity estimation within the time period in which the inverse power condition is determined. Active power injection at grid connection point The direction is used as the criterion, and the following formula is used to determine the reverse power condition: in, This is a marker for reverse power operation. The inverse power discrimination threshold, It is an indicator function; Within the inverse power condition sample, a perturbation sample sequence is constructed using differential methods: in, For grid connection point voltage disturbance, and The disturbances are injected into the active and reactive power respectively; within the rolling window, sample points that meet the disturbance amplitude threshold are selected as valid disturbance samples.

Images