Data-driven power distribution network accommodation capacity evaluation method and system

Abstract

This invention belongs to the field of distribution network planning technology, specifically involving a data-driven method and system for assessing the acceptance capacity of distribution networks. The method includes: collecting historical time-series measurement data from each node of the distribution network; calculating instantaneous voltage sensitivity and its rate of change using the ratio of measured voltage change to measured power change; constructing a dual-coupling weight by combining the deviation between voltage amplitude and the voltage safety upper limit; constructing an improved overall least-squares objective function; weighting the objective function across all time periods using the dual-coupling weight; identifying the dynamic voltage sensitivity sequence under physical constraints; calculating the fitting residual and its standard deviation based on the identified dynamic voltage sensitivity sequence; constructing a global confidence upper limit for sensitivity; and calculating the acceptance capacity of the distribution network by combining the voltage safety upper limit and the current operating voltage. This invention achieves robust assessment of acceptance capacity under unknown parameters.

Description

Technical Field

[0001] This invention relates to the field of distribution network planning technology. More specifically, this invention relates to a data-driven method and system for assessing the acceptance capacity of distribution networks. Background Technology

[0002] With the advancement of energy transition and the exponential growth in the penetration rate of new power sources such as distributed photovoltaics and electric vehicle charging stations in distribution networks, the distribution network is transforming from a traditional unidirectional radial network into an active network with interaction between power sources and loads. Against this backdrop, accurately assessing the capacity of each node in the distribution network to accept distributed power sources—that is, the maximum power allowed to be connected without violating safety constraints such as voltage and thermal stability—has become crucial for ensuring the safe operation of the power grid and guiding new energy planning.

[0003] Traditional physical model-based evaluation methods rely on accurate topology and line impedance parameters. However, medium and low voltage distribution networks often suffer from unclear topology, missing or drifting physical parameters due to their age or aging lines. Under the black box condition of unknown physical parameters, traditional methods are difficult to implement because they cannot establish accurate power flow equations, or the evaluation results deviate significantly from reality due to parameter errors.

[0004] Data-driven methods use historical measurement data to identify the relationship between voltage and power. However, the commonly used least squares method usually assumes that the injected power is accurate and ignores the measurement noise inherent in the power readings of smart meters. In addition, existing weighting strategies only assign static weights based on voltage amplitude, ignoring the significant nonlinear saturation characteristics of the power-voltage characteristic curve when the distribution network is close to its acceptance limit. This makes it difficult to accurately pinpoint the critical physical boundary of voltage over-limit, resulting in the evaluation model failing to balance safety and resource utilization at key over-limit boundaries. Summary of the Invention

[0005] To address the technical problems of existing distribution network acceptance capacity assessment methods, such as optimization deviations due to inconsistencies in voltage and power dimensions and failure to capture nonlinear characteristics of critical states, which exist under conditions of missing physical parameters and bidirectional measurement noise interference, this invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a data-driven method for assessing the acceptance capacity of a distribution network, comprising:

[0007] Historical time-series measurement data of each node in the distribution network are collected, and the changes in measured power and voltage are calculated. The approximate instantaneous voltage sensitivity and its rate of change are calculated using the ratio of the measured voltage change to the measured power change. A dual-coupling weight is constructed by combining the deviation between the voltage amplitude and the voltage safety upper limit. An improved overall least squares objective function is constructed, which includes terms representing the deviation between the measured power change and its true value, as well as terms representing the deviation between the measured voltage change and its true value. The dual-coupling weight is used to weight the objective function across all time periods, and the dynamic voltage sensitivity sequence is identified under physical constraints. Based on the identified dynamic voltage sensitivity sequence, the fitting residual and its standard deviation are calculated to construct a global confidence upper limit for voltage sensitivity. The acceptance capacity of the distribution network is then calculated by combining the voltage safety upper limit with the current operating voltage.

[0008] This invention provides a dynamic incremental basis for data-driven analysis by collecting historical time-series measurement data and calculating the changes in power and voltage, thus removing static biases. It calculates instantaneous voltage sensitivity and its rate of change using the ratio of voltage to power changes, and constructs a dual-coupling weight based on the deviation between voltage amplitude and the safety upper limit. This simultaneously captures the voltage exceedance risk of the distribution network under critical conditions and the nonlinear morphological changes of the power-voltage characteristic curve. An improved overall least-squares objective function is constructed and weighted using the dual-coupling weight. Under physical constraints, dynamic voltage sensitivity is identified, eliminating optimization bias caused by the inconsistency between power and voltage dimensions and overcoming the influence of two-sided measurement noise. This ensures that the identification results conform to physical laws and focus on high-risk operating points. Based on the identification results, a global confidence upper limit is constructed and acceptance capability is calculated, transforming model uncertainty into a safety margin, thereby achieving robust evaluation under unknown parameters and data noise interference.

[0009] Preferably, the calculation of the measured power change and the measured voltage change includes: smoothing the active power sequence and voltage amplitude sequence in the historical time series measurement data respectively, calculating the time difference between the smoothed active power sequence and voltage amplitude sequence, and obtaining the measured power change and the measured voltage change at each moment.

[0010] Preferably, the dual coupling weights satisfy the expression: In the formula, Indicates the first Dual coupling weights at each sampling time; Represents the normalization function; Indicates the first Voltage amplitude at each sampling time; Indicates the upper limit of voltage safety; Represents an exponential function with the natural constant as its base; This represents the preset risk scaling parameters; Indicates the first The change in the approximate instantaneous voltage sensitivity at each sampling moment; This represents the long-term average absolute rate of change of the approximate value of voltage sensitivity. Indicates the first The measured power change at each sampling time; This represents the standard deviation of the measured power change. This represents a small positive constant to prevent the denominator from being zero; its empirical value is... ; This represents the curvature sensitivity coefficient.

[0011] This invention constructs a comprehensive index that simultaneously reflects the urgency of voltage safety and the hardening of the power grid's physical form by integrating the deviation of voltage amplitude from the safety upper limit and the rate of change of the instantaneous voltage sensitivity approximation. It utilizes an exponential function to nonlinearly amplify the importance of samples approaching the voltage upper limit, combines the rate of change of sensitivity to capture the bending saturation characteristics of the power-voltage curve, and unifies the scale through a normalization function. This enables the subsequent parameter identification process to automatically focus on the most critical harsh operating condition samples for assessing acceptance capability, significantly improving the assessment accuracy under critical conditions.

[0012] Preferably, the overall least squares objective function satisfies the expression:

[0013] ;

[0014] In the formula, This represents the full-time dynamic voltage sensitivity sequence to be solved; Indicates the first Voltage sensitivity at each sampling time; Indicates the first The actual power change after correction at each sampling time; Indicates the first The actual voltage change after correction at each sampling time; Indicates the first The measured power change at each sampling time; Indicates the first The measured voltage change at each sampling time; This represents the variance of the measured power change. This represents the variance of the measured voltage change. Indicates the first Dual coupling weights at each sampling time; This represents the total number of samples in historical time-series measurement data. Represents the square of the Euclidean norm; Represents the smoothing regularization coefficient; This represents a typical value indicating voltage sensitivity.

[0015] This invention introduces the reciprocals of the variances of measured power change and measured voltage change as weighting factors, thereby achieving dimensionless residual terms in the objective function. This eliminates the optimization direction deviation caused by the large differences in the magnitudes of power and voltage values, ensuring balanced processing of measurement errors on both sides. Simultaneously, the introduction of a smoothing regularization term utilizes prior knowledge of the short-term invariance of the distribution network topology, effectively suppressing non-physical abrupt changes in sensitivity caused by data fluctuations. Combined with the weighting effect of dual-coupling weights, the identified full-time dynamic voltage sensitivity sequence is optimized in terms of noise resistance, smoothness, and key point fitting accuracy.

[0016] Preferably, identifying the dynamic voltage sensitivity sequence under applied physical constraints includes: satisfying the constraint conditions. and Under the premise of minimizing the overall least squares objective function, solve for the dynamic voltage sensitivity sequence, where, This is the lower physical bound.

[0017] Preferably, the fitting residual satisfies the expression: In the formula, Indicates the first Voltage fitting residuals at each sampling time; Indicates the first The measured voltage change at each sampling time; The first one obtained by identification Voltage sensitivity at each sampling time; Indicates the first The measured power change at each sampling time.

[0018] Preferably, the construction of the global confidence upper limit for voltage sensitivity includes: selecting the top 5% of samples with the largest values ​​in the dynamic voltage sensitivity sequence over the entire time period, calculating their mean as the average voltage sensitivity of the high-risk interval; and constructing the global confidence upper limit based on the average voltage sensitivity of the high-risk interval and the standard deviation of the fitted residual sequence.

[0019] This invention selects the samples with the largest values ​​in the dynamic voltage sensitivity sequence over the entire time period and calculates their mean as the average voltage sensitivity in the high-risk range, thereby locking in the benchmark sensitivity level of the power grid under the worst operating conditions and avoiding the masking of risks by averaging.

[0020] Preferably, the global confidence upper limit satisfies the expression: In the formula, Represents the global confidence upper limit for voltage sensitivity; This indicates the average voltage sensitivity in the high-risk range; For statistical confidence factors; This represents the standard deviation of the fitted residual sequence.

[0021] This invention achieves probabilistic coverage of the true sensitivity value by superimposing an error margin composed of statistical confidence factors and the standard deviation of the fitting residuals on the average voltage sensitivity in the high-risk range that reflects harsh operating conditions. This transforms the uncertainty of the model into a specific numerical boundary, ensuring that the true voltage sensitivity will not exceed this upper limit in the vast majority of cases. This prevents the risk of overestimating acceptance capability due to underestimating sensitivity and ensures the safety of the evaluation results.

[0022] Preferably, the acceptance capacity of the distribution network satisfies the expression: In the formula, This indicates the ability to accept; Indicates the upper limit of voltage safety; This indicates the operating voltage at the current moment; Represents the global confidence upper limit for voltage sensitivity; For statistical coverage factor; This represents the standard deviation of the measured power change. This represents the function that takes the maximum value.

[0023] This invention transforms voltage constraints into power constraints by dividing the difference between the upper limit of voltage safety and the current operating voltage by the global confidence upper limit of voltage sensitivity. Furthermore, it subtracts a fluctuation reserve term composed of statistical coverage factor and standard deviation of measured power change from this. This not only considers the steady-state carrying capacity of the power grid under the worst sensitivity but also reserves a dynamic safety margin to cope with instantaneous power changes in source load. The non-negativity of the result is ensured by taking the maximum value function. The final result is a conservative and robust acceptance capacity value that can make full use of the existing capacity of the power grid while strictly preventing voltage overruns.

[0024] Secondly, the present invention provides a data-driven distribution network accessibility assessment system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned data-driven distribution network accessibility assessment method is implemented.

[0025] By adopting the above technical solution, the data-driven distribution network acceptance capacity assessment method is generated into a computer program and stored in a memory for loading and execution by a processor. Terminal equipment is then manufactured based on the memory and processor for convenient use.

[0026] The beneficial effects of this invention are as follows: This invention constructs a dynamic analysis foundation by collecting historical time-series measurement data from each node of the distribution network and calculating the measured power change and measured voltage change; it calculates the instantaneous voltage sensitivity and its rate of change using the ratio of the measured voltage change to the measured power change, and constructs a dual-coupling weight by combining the deviation between the voltage amplitude and the voltage safety upper limit. This simultaneously captures the voltage exceedance risk of the distribution network under critical conditions and the nonlinear morphological changes of the power-voltage characteristic curve, solving the problem that traditional static weighting cannot adapt to nonlinear characteristics; this invention constructs an improved overall least squares objective function. By weighting the deviation term using the variances of measured power changes and measured voltage changes to eliminate dimensional differences, and by using dual-coupling weights to weight the objective function across the entire time period, dynamic voltage sensitivity sequences are identified under physical constraints. This not only overcomes the influence of two-sided measurement noise but also ensures that the identification results conform to physical laws and focus on high-risk operating points. Based on the identified dynamic voltage sensitivity sequences, this invention constructs a global confidence upper limit for voltage sensitivity and calculates acceptance capability, transforming the uncertainty of the model into a safety margin, thereby achieving robust assessment of distribution network acceptance capability under unknown parameters and data noise interference. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating the data-driven distribution network acceptance capacity assessment method of the present invention;

[0028] Figure 2 This is a schematic diagram illustrating the distribution of dual coupling weights on the power-voltage plane;

[0029] Figure 3 This is a schematic diagram illustrating the dynamic voltage sensitivity curve obtained from the identification.

[0030] Figure 4 This is a schematic diagram illustrating the acceptance capacity curve. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0033] This invention discloses a data-driven method for assessing the acceptance capacity of distribution networks, referring to... Figure 1This includes steps S1-S4:

[0034] S1. Collect historical time-series measurement data of each node in the distribution network, and calculate the changes in measured power and measured voltage.

[0035] It should be noted that the operating data of the distribution network includes information in two dimensions: voltage and power. These two dimensions differ fundamentally in numerical magnitude and physical units. Furthermore, the measured data often contains static biases. If the raw data is used directly for analysis, the biases at static operating points will mask the dynamic changes. Therefore, this invention first performs differential processing on historical data to eliminate static biases, focuses on the physical coupling relationship between increments, and calculates the statistical variance of the distribution network operating data. This provides a normalized benchmark for subsequently converting physical quantities of different dimensions into dimensionless signal-to-noise ratios.

[0036] Specifically, the active power sequence and voltage amplitude sequence of the node to be evaluated in the distribution network are collected within a preset time window. These sequences constitute the historical time-series measurement data of the node. The active power sequence and voltage amplitude sequence are then smoothed to filter out high-frequency random noise. The time difference between the smoothed active power sequence and voltage amplitude sequence is calculated to obtain the measured power change at each moment within the time window. and the amount of voltage change ,in, Indicates the first The measured power change at each sampling time. , They represent the first The sampling time and the first sampling time Active power at each sampling time. Indicates the first The measured voltage change at each sampling time. , They represent the first The sampling time and the first sampling time Voltage amplitude at each sampling time.

[0037] It should be noted that the present invention uses the moving average method to smooth the active power sequence and voltage amplitude sequence. In other embodiments, implementers may also choose other algorithms to smooth the active power sequence and voltage amplitude sequence, such as the Savitzky-Golay filter.

[0038] Furthermore, the variance of the measured power variation over the entire time period is calculated. and the variance of the measured voltage change And calculate the standard deviation of the measured power change. This serves as a scaling factor for the subsequent construction of the dimensionless objective function and the calculation of volatility reserves.

[0039] S2. Calculate the approximate value of instantaneous voltage sensitivity and its rate of change using the ratio of the measured voltage change to the measured power change. Combine this with the deviation between the voltage amplitude and the upper limit of voltage safety to construct a dual coupling weight.

[0040] It is important to note that the core of capacity assessment lies in accurately capturing the grid's response under pressure limits. When the distribution network approaches its voltage limit, the nonlinear characteristics of lines and transformers cause the active power voltage characteristic curve to exhibit bending saturation characteristics, meaning that the voltage sensitivity to power changes drastically. Existing technologies typically only consider the distance of the voltage amplitude from the upper limit, ignoring the rate of dynamic morphological deterioration of the system. Therefore, this invention introduces the concepts of instantaneous voltage sensitivity and its rate of change, constructing a dual-coupling weight to simultaneously capture the risk of voltage approaching the limit and the abrupt change in sensitivity, thereby achieving focus on key physical samples at the algorithmic level.

[0041] Specifically, define the first The approximate instantaneous voltage sensitivity at each sampling time is: And calculate its time difference. ,in, Indicates the first The measured voltage change at each sampling time. Indicates the first The measured power change at each sampling time. Indicates the first The change in the approximate voltage sensitivity value at each sampling time. Indicates the first The approximate instantaneous voltage sensitivity value at each sampling time. It should be noted that the present invention uses the ratio of the measured voltage change to the measured power change as an approximate instantaneous voltage sensitivity value, which can reflect the sensitivity of the power grid to power fluctuations at the current operating point, i.e., the instantaneous stiffness characteristic.

[0042] Furthermore, the dual coupling weights at the sampling time are calculated:

[0043]

[0044] In the formula, Indicates the first Dual coupling weights at each sampling time; This represents a normalization function used to map the calculation result to the interval between 0 and 1. In this embodiment, maximum and minimum value normalization is used. Indicates the first Voltage amplitude at each sampling time; Indicates the upper limit of voltage safety; Represents an exponential function with the natural constant as its base; This represents a risk scaling parameter used to adjust the nonlinear sensitivity to voltage over-limit trends. Indicates the first The change in the approximate instantaneous voltage sensitivity at each sampling moment; This represents the long-term average absolute rate of change of the approximate value of voltage sensitivity, i.e. Used to eliminate Dimensions This represents the total number of samples in historical time-series measurement data. Indicates the first The measured power change at each sampling time; The standard deviation of the measured power change is used to eliminate... Dimensions; This represents a small positive constant to prevent the denominator from being zero; its empirical value is... The implementers can also set it according to the actual implementation situation, but in order to ensure the stability of the numerical calculation, The range of values ​​is to between; This represents the curvature sensitivity coefficient, which is used to amplify the voltage sensitivity rate of change characteristic in the total weight.

[0045] In the formula, Reflecting the risk of voltage exceeding limits, this invention utilizes the nonlinear amplification characteristics of an exponential function, when the voltage amplitude... Less than the voltage safety limit and approaching hour, A sharp increase indicates a higher level of risk, when the voltage amplitude... Greater than the voltage safety limit At this point, the weights are further increased, forcing the model to focus on samples that exceed the limits.

[0046] In the formula, It reflects the rate of change of normalized voltage sensitivity relative to power, corresponding to the second derivative characteristic of the power-voltage characteristic curve, i.e., curvature. When the value of this term is larger, it indicates that the voltage sensitivity to power is changing drastically, and the power grid is in the nonlinear hardening stage.

[0047] Risk scaling parameters The risk scaling parameter determines the model's nonlinear sensitivity to voltage limit exceedance trends. The decay rate of the exponential function can be controlled by the risk scaling parameter, thus determining the steepness of the risk weights in the critical region. This embodiment uses the risk scaling parameter... In other embodiments, the implementer can select the specific value of the risk scaling parameter based on the voltage level and operating standards of the actual power grid. For example, if the voltage quality requirements in the area to be evaluated are extremely high and there are precision loads that are highly sensitive to voltage fluctuations, the penalty for voltage deviation needs to be increased, and the value of the risk scaling parameter can be reduced. If the area to be evaluated is a rural distribution network with strong voltage tolerance, a certain range of short-term fluctuations needs to be tolerated, and the value of the low-risk scaling parameter can be increased.

[0048] Curvature sensitivity coefficient The amplification factor of the voltage sensitivity change rate characteristic in the total weight is determined. In this embodiment, the curvature sensitivity coefficient is set to 3. In other embodiments, the implementer can select the specific value of the curvature sensitivity coefficient according to the actual implementation situation. For example, if the focus is on capturing weak precursor signals before the power grid collapse, it is necessary to improve the ability to capture morphological changes, so the value of the curvature sensitivity coefficient can be increased. If the focus is on obtaining the average characteristics over the entire time period, it is necessary to reduce the impact of local changes on the overall fitting, so the value of the curvature sensitivity coefficient can be decreased.

[0049] For example, Figure 2 The distribution of dual coupling weights on the power-voltage plane is shown in the figure. The color intensity of the scatter points in the figure represents the magnitude of the dual coupling weights. The brighter the color, the greater the dual coupling weight, and the darker the color, the smaller the dual coupling weight. When the voltage amplitude approaches the safety limit, the color of the scatter points becomes significantly brighter, indicating that the dual coupling weight increases sharply. The dual coupling weight can effectively identify and focus on key physical samples with high risk of voltage exceeding the limit, thereby giving these samples a higher fitting priority in subsequent optimization.

[0050] S3. Construct an improved overall least squares objective function, which includes a deviation term between the measured power change and its true value and a deviation term between the measured voltage change and its true value. Use dual coupling weights to weight the objective function for the entire time period and identify the dynamic voltage sensitivity sequence under physical constraints.

[0051] It should be noted that accurately identifying voltage sensitivity under black-box conditions with unknown parameters requires addressing two core issues: first, the inconsistency between the dimensions of power and voltage leads to confusion in the physical meaning of the optimization objective; and second, data noise may cause the identification results to violate the physical laws of the power grid. Therefore, this invention improves upon the traditional total least squares method by introducing the inverse of variance as a weighting factor to transform the residuals into a dimensionless form, and applying non-negative physical constraints and smoothing regularization.

[0052] Specifically, an improved overall least squares optimization model is constructed to solve the dynamic voltage sensitivity sequence over the entire time period. It satisfies the expression:

[0053]

[0054]

[0055] In the formula, This represents the full-time dynamic voltage sensitivity sequence to be solved; Indicates the first Dynamic voltage sensitivity at each sampling time; Indicates the first The actual power change after correction at each sampling time; Indicates the first The actual voltage change after correction at each sampling time; Indicates the first The measured power change at each sampling time; Indicates the first The measured voltage change at each sampling time; This represents the variance of the measured power change. This represents the variance of the measured voltage change. Indicates the first Dual coupling weights at each sampling time; This represents the total number of samples in historical time-series measurement data. Represents the square of the Euclidean norm; Represents the smoothing regularization coefficient; This represents a typical value of voltage sensitivity, taking the average of the absolute values ​​of the approximate instantaneous voltage sensitivity, used to make the regularization term dimensionless; Indicates the lower physical bound; This represents the combination of variables that minimizes the objective function.

[0056] In the formula, by multiplying by and This invention standardizes the power residual and voltage residual terms to a relative fluctuation level, eliminating the unit difference between kilowatts and volts, and enabling the optimization process to fairly balance the measurement noise of the two dimensions. This measure utilizes the prior knowledge that the physical topology of the distribution network remains unchanged in the short term to constrain the voltage sensitivity at adjacent time points from undergoing drastic, non-physical jumps; constraint conditions This ensures that the voltage sensitivity is always positive and conforms to the common sense of physics, eliminating pseudo-physical solutions caused by data noise.

[0057] Smoothing regularization coefficient The strength of the continuity constraint on the identification results in the time dimension is determined. In this embodiment, the smoothing regularization coefficient is used. In other embodiments, the smoothing regularization coefficient can be set to 15 according to the accuracy level of the measurement device. For example, if the signal-to-noise ratio of the measurement data is low and there is a lot of random noise interference, the smoothing constraint needs to be enhanced to suppress pseudo fluctuations, and a high value close to 20.0 can be taken; if there are frequent topology switching or fast adjustment devices in the distribution network, the fast time-varying characteristics of the parameters need to be preserved, and a low value close to 10.0 can be taken.

[0058] Physical lower bound The physical effective domain boundary of the identification result is determined to prevent the optimization algorithm from falling into zero or negative value traps that violate circuit principles during the numerical solution process. In this embodiment, the physical lower bound is defined as follows: Set to 0.05. In other embodiments, implementers may estimate the physical lower bound based on line parameter files.

[0059] S4. Based on the identified dynamic voltage sensitivity sequence, calculate the fitting residual and its standard deviation, construct the global confidence upper limit of voltage sensitivity, and combine the voltage safety upper limit with the current operating voltage to calculate the acceptance capacity of the distribution network.

[0060] It should be noted that since any parameter identification has residual error, a single value of the acceptance capacity cannot reflect the risk brought about by the estimation error. In order to ensure the absolute safety of the power grid operation, this invention does not directly use the identified dynamic voltage sensitivity estimate. Instead, it constructs a confidence interval based on statistical principles and selects the statistical upper bound of the high-risk interval as the evaluation benchmark to calculate the conservative lower bound of the acceptance capacity, thus transforming uncertainty into a safety margin.

[0061] Specifically, based on the identified dynamic voltage sensitivity sequence Calculate the fitted residual sequence , of which The residual at each moment Satisfying the expression:

[0062]

[0063] In the formula, Indicates the first The voltage fitting residual at each sampling time reflects the deviation between the identification model and the actual measurement data; Indicates the first The measured voltage change at each sampling time; The first one obtained by identification Dynamic voltage sensitivity at each sampling time; Indicates the first The measured power change at each sampling time.

[0064] Furthermore, the standard deviation of the fitted residual sequence is calculated. The top 5% of samples with the highest values ​​in the dynamic voltage sensitivity sequence over the entire time period were selected, and their mean was calculated as the average voltage sensitivity in the high-risk range. A global confidence upper limit for voltage sensitivity is constructed based on the standard deviation of the fitted residual sequence and the average voltage sensitivity in the high-risk region:

[0065]

[0066] In the formula, Represents the global confidence upper limit for voltage sensitivity; It represents the average voltage sensitivity in the high-risk range and is used to characterize the reference voltage sensitivity level of the power grid under adverse operating conditions; In this embodiment, the statistical confidence factor is used. This corresponds to a confidence level of approximately 95%, used to cover random errors in parameter identification. In other embodiments, implementers can set the statistical confidence factor according to the actual implementation situation. ; Represents the standard deviation of the fitted residual sequence, when The larger the value, the more vulnerable the power grid is statistically, and the more conservative the calculated acceptance capacity will be, thus ensuring the security of the results.

[0067] For example, Figure 3 The dynamic voltage sensitivity identified by this invention shows that the approximate instantaneous voltage sensitivity is affected by measurement noise, fluctuates drastically, and contains a large number of glitches. However, the dynamic voltage sensitivity identified by this invention filters out noise interference while retaining the physical trend of sensitivity changing over time, and the curve is smooth and continuous, which conforms to the continuity characteristics of the physical state of the distribution network.

[0068] Further, calculate the current node's acceptance capacity:

[0069]

[0070] In the formula, This indicates the capacity to accept data, reflecting the maximum accessible power after taking robustness risks into account. Indicates the upper limit of voltage safety; This indicates the operating voltage at the current moment; Represents the global confidence upper limit for voltage sensitivity; In this embodiment, the statistical coverage factor is used. This is used to reflect the 3-standard-deviation principle, which means reserving the voltage safety margin required to cope with most instantaneous power surges; This represents the standard deviation of the measured power change. This constitutes a fluctuation reserve to prevent voltage overshoot caused by sudden changes in photovoltaic power output; This represents the function that takes the maximum value, ensuring that the calculated acceptance capacity is not negative.

[0071] For example, Figure 4 The red arrows mark the bottleneck moments and corresponding minimum values ​​of the acceptance capacity curve throughout the entire time period, indicating that the present invention can dynamically measure the acceptance capacity based on the real-time operating status of the power grid and provide the safety baseline under the worst operating conditions, thus providing a decision-making basis for the dispatch and operation of the distribution network.

[0072] This invention also discloses a data-driven distribution network accessibility assessment system, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the data-driven distribution network accessibility assessment method according to this invention.

[0073] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A data-driven method for assessing the acceptance capacity of distribution networks, characterized in that, include: Collect historical time-series measurement data of each node in the distribution network, and calculate the changes in measured power and measured voltage. An approximate value for instantaneous voltage sensitivity and its rate of change are calculated using the ratio of the measured voltage change to the measured power change. A dual-coupling weighting is then constructed by combining the deviation between the voltage amplitude and the upper limit of voltage safety. In the formula, Indicates the first Dual coupling weights at each sampling time; Represents the normalization function; Indicates the first Voltage amplitude at each sampling time; Indicates the upper limit of voltage safety; Represents an exponential function with the natural constant as its base; This represents the preset risk scaling parameters; Indicates the first The change in the approximate instantaneous voltage sensitivity at each sampling moment; This represents the long-term average absolute rate of change of the approximate value of voltage sensitivity. Indicates the first The measured power change at each sampling time; This represents the standard deviation of the measured power change. This represents a small positive constant to prevent the denominator from being zero; its empirical value is... ; Indicates the curvature sensitivity coefficient; Construct an improved overall least squares objective function. ; In the formula, This represents the full-time dynamic voltage sensitivity sequence to be solved; Indicates the first Voltage sensitivity at each sampling time; Indicates the first The actual power change after correction at each sampling time; Indicates the first The actual voltage change after correction at each sampling time; Indicates the first The measured voltage change at each sampling time; This represents the variance of the measured power change. This represents the variance of the measured voltage change. This represents the total number of samples in historical time-series measurement data. Represents the square of the Euclidean norm; Represents the smoothing regularization coefficient; The objective function, which represents a typical value of voltage sensitivity, includes a term representing the deviation between the measured power change and its true value, and a term representing the deviation between the measured voltage change and its true value. The objective function is weighted over the entire time period using dual coupling weights, and the dynamic voltage sensitivity sequence is identified under the condition of applying physical constraints. Based on the identified dynamic voltage sensitivity sequence, the fitting residual and its standard deviation are calculated, a global confidence upper limit for voltage sensitivity is constructed, and the acceptance capacity of the distribution network is calculated by combining the voltage safety upper limit with the current operating voltage.

2. The data-driven distribution network acceptance capacity assessment method according to claim 1, characterized in that, The calculation of the measured power change and the measured voltage change includes: The active power sequence and voltage amplitude sequence in the historical time series measurement data are smoothed respectively. The time difference between the smoothed active power sequence and voltage amplitude sequence is calculated to obtain the measured power change and measured voltage change at each moment.

3. The data-driven distribution network acceptance capacity assessment method according to claim 1, characterized in that, The identification of dynamic voltage sensitivity sequences under applied physical constraints includes: Under the condition of satisfying the constraints and Under the premise of minimizing the overall least squares objective function, solve for the dynamic voltage sensitivity sequence, where, This is the lower physical bound.

4. The data-driven distribution network acceptance capacity assessment method according to claim 1, characterized in that, The fitting residual satisfies the expression: ； In the formula, Indicates the first Voltage fitting residuals at each sampling time; Indicates the first The measured voltage change at each sampling time; The first one obtained by identification Dynamic voltage sensitivity at each sampling time; Indicates the first The measured power change at each sampling time.

5. The data-driven distribution network acceptance capacity assessment method according to claim 1, characterized in that, The global confidence upper limit for constructing voltage sensitivity includes: The top 5% of samples with the largest values ​​in the dynamic voltage sensitivity sequence across all time periods are selected, and their mean is calculated as the average voltage sensitivity in the high-risk interval. Based on the average voltage sensitivity in the high-risk interval and the standard deviation of the fitted residual sequence, a global confidence upper limit is constructed.

6. The data-driven distribution network acceptance capacity assessment method according to claim 5, characterized in that, The global confidence upper limit satisfies the expression: ； In the formula, Represents the global confidence upper limit for voltage sensitivity; This indicates the average voltage sensitivity in the high-risk range; For statistical confidence factors; This represents the standard deviation of the fitted residual sequence.

7. The data-driven distribution network acceptance capacity assessment method according to claim 1, characterized in that, The acceptance capacity of the distribution network satisfies the following expression: ； In the formula, This indicates the ability to accept; Indicates the upper limit of voltage safety; This indicates the operating voltage at the current moment; Represents the global confidence upper limit for voltage sensitivity; For statistical coverage factor; This represents the standard deviation of the measured power change. This represents the function that takes the maximum value.

8. A data-driven distribution network acceptance capacity assessment system, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the data-driven distribution network access capability assessment method according to any one of claims 1-7.

Images