A photovoltaic power consumption boundary evaluation method and system for a low-voltage alternating-direct current hybrid distribution area

Abstract

The application belongs to the technical field of distribution network photovoltaic evaluation, and discloses a photovoltaic consumption boundary evaluation method and system for a low-voltage AC-DC hybrid transformer area, wherein through port equivalence and hierarchical division of an AC-DC full node, combined with end homogeneous node Thevenin aggregation reduction processing, the bidirectional converter is equivalent to mutual admittance interaction ports and is configured with an adaptive phase shift correction coefficient, a unified impedance network model covering the full node is constructed, and the voltage coupling characteristics of the photovoltaic output fluctuation on the DC side through the converter to the AC side are restored; through multi-working condition sampling and quadratic surface fitting, a converter efficiency mapping function is constructed, the total loss is split into series conduction loss and parallel switch loss and is equivalent to a variable impedance embedded in power flow iteration, in each round of calculation, the impedance parameters are updated synchronously and the power deviation is corrected reversely, and the modulation effect of transmission power proportion, environmental temperature and DC bus voltage on conversion efficiency is dynamically adapted.

Description

Technical Field

[0001] This invention belongs to the field of distribution network photovoltaic assessment technology, specifically a method and system for assessing the photovoltaic absorption boundary of a low-voltage AC / DC hybrid distribution area. Background Technology

[0002] Low-voltage AC / DC hybrid distribution areas have become an important development form of new power distribution systems due to their flexible power interaction and low line losses. Photovoltaic (PV) absorption boundary assessment is a core technical support for guiding the orderly integration of PV power into distribution areas, mitigating grid operation risks, and improving the utilization rate of new energy sources. While existing technologies have yielded considerable research results on PV absorption assessment for pure AC distribution areas, assessment methods for AC / DC hybrid distribution areas still face the following technical challenges: Existing assessments generally adopt a decoupled calculation mode for AC / DC subgrids, calculating the absorption capacity of both sides separately and then directly superimposing them, ignoring the power interaction loss of the bidirectional converter and the coupling and conduction effect of the voltage of the two sides, resulting in a significant deviation between the assessment results and the actual operating conditions; Most methods assume that the converter operates at a constant rated efficiency, without considering the dynamic modulation effect of transmission power ratio, ambient temperature, and DC bus voltage fluctuations on the converter conversion efficiency. In scenarios where photovoltaic output fluctuates frequently throughout the day, the calculation results of the absorption boundary show significant time-series drift, and the evaluation error directions are opposite for peak and trough periods. The existing absorption constraint system mostly covers only two types of conventional electrical constraints: node voltage over-limit and branch current carrying capacity. It does not include implicit operational constraints such as the residual current protection action threshold of the transformer area and the DC side ripple voltage suppression requirements in the evaluation. This makes it easy for the calculated theoretical absorption boundary to trigger the protection device to malfunction in actual operation. The standard parameters of the line are generally used for deterministic assessment, without taking into account the impedance aging deviation of old cables in low-voltage distribution areas and the parameter dispersion caused by construction technology. The assessment results lack a confidence reference dimension, which leads to either being too conservative and wasting space for absorption or being too aggressive and creating hidden operational risks in engineering applications. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for assessing the photovoltaic absorption boundary of low-voltage AC / DC hybrid distribution areas, so as to solve one or more problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a photovoltaic absorption boundary assessment method for low-voltage AC / DC hybrid distribution areas, comprising a unified impedance modeling stage, a loss equivalent embedding stage, a constraint system construction stage, a typical scenario generation stage, an initial boundary solution stage, a confidence boundary correction stage, and a boundary verification output stage. Furthermore, in the unified impedance modeling stage, the AC distribution cables, DC busbars, bidirectional converters, photovoltaic grid-connected nodes, and AC / DC load nodes of the target transformer area are port equivalent. On the AC side, nodes are divided step by step according to the low-voltage outgoing terminals of the distribution transformer, branch box nodes, and end user nodes. On the DC side, nodes are divided according to the photovoltaic combiner port, energy storage access port, and DC load port. The node numbering adopts continuous integer encoding, and AC nodes and DC nodes are numbered in intervals. Thevenin equivalent aggregation is performed on multiple end nodes with the same electrical characteristics under the same distribution branch, and the end user node is equivalent to a single aggregation port. The constant power external characteristics of the photovoltaic grid-connected unit, the droop control characteristics of the energy storage unit, and the impedance characteristics of the DC load are embedded into the self-admittance terms of the corresponding nodes in the form of admittance increments. At the same time, an adaptive phase shift correction coefficient is set for the mutual admittance coupling terms of the bidirectional converter, and the aggregated AC subgrid node admittance matrix and DC subgrid node admittance matrix are constructed respectively. The bidirectional converter is equivalent to the mutual admittance interaction port connecting the AC and DC subnets. Combining the power transmission characteristics of the converter in both rectification and inversion modes, the converter port voltage ratio and the corrected phase shift characteristics are transformed into coupling terms in the admittance matrix, ultimately forming a unified impedance network model that includes all AC and DC nodes.

[0005] AC / DC unified impedance network node voltage equations:

[0006] This represents the admittance matrix of the AC subnet nodes after Thevenin aggregation and order reduction at the end homogeneous nodes. The matrix elements include the inherent admittance parameters of AC power distribution cables, as well as the admittance increments equivalent to the operating characteristics of AC loads and grid-connected equipment. This represents the admittance matrix of DC subnet nodes. The matrix elements include the inherent admittance parameters of DC buses and DC lines, as well as the equivalent admittance increments of photovoltaic, energy storage, and DC load operating characteristics. , The mutual admittance coupling matrix of the bidirectional converter is the interaction port coupling term of the AC / DC subnet. It has a built-in adaptive phase shift correction coefficient, which can smoothly adapt to the rectification and inverter bidirectional power transmission conditions. This represents the column vector of voltage phasors for all electrical nodes on the AC side. This represents a column vector of voltage magnitudes for all electrical nodes on the DC side. This represents the column vector of injected current phasors for all electrical nodes on the AC side. This represents the column vector of injected current magnitudes for all electrical nodes on the DC side.

[0007] Furthermore, in the loss equivalent embedding stage, the conversion efficiency data of the converter under different transmission power ratios, ambient temperatures, and DC bus voltages are collected through bench testing or simulation. The transmission power ratio is sampled at 5% intervals in the 0-10% light load range and traversed at 10% intervals in the 10% to rated power range. The ambient temperature covers the typical operating range of the transformer area from -10℃ to 45℃. The least squares method is used to fit the quadratic surface. The fitting process takes the minimum mean square error as the objective function. For outliers, a weighted regression method is used to reduce the corresponding weights, thereby weakening the interference of outlier data on the fitting results. The coupling functions of converter efficiency, power, temperature and voltage are obtained. The efficiency of a bidirectional converter is mapped to a quadratic surface function:

[0008] The real-time conversion efficiency of the bidirectional converter is represented by the ratio of active power on the output side to active power on the input side, with a value range of 0 to 1. A unified efficiency calculation logic is used under both rectification and inversion conditions, and it is not affected by the power flow direction, but is determined only by the current operating condition parameters.

[0009] This represents the transmission power ratio of the bidirectional converter, i.e., the ratio of the current transmission power amplitude to the converter's rated active power, with a value range of 0 to 1. In the actual measurement sampling settings, the 0-10% light load range is sampled at 5% intervals, and the 10% to rated power range is traversed at 10% intervals.

[0010] This indicates the ambient temperature at which the converter operates, expressed in °C. It covers the typical operating range of the transformer area from -10 °C to 45 °C. Temperature changes alter the on-state voltage drop and switching characteristics of power devices, thereby modulating the conversion efficiency.

[0011] This represents the per-unit value of the DC bus voltage of the converter. It is normalized based on the rated DC bus voltage. Fluctuations in the DC bus voltage directly change the switching loss level of the power devices.

[0012] ~ The 10 undetermined coefficients for the quadratic surface fitting are obtained by iterative solution using the weighted least squares method. The fitting process takes minimizing the mean square error as the objective function, and the weights of outliers with fitting residuals exceeding 3 times the mean square error are reduced. At the same time, the baseline weights of the sampling points are assigned according to the actual occurrence probability of the corresponding operating conditions in the transformer area throughout the year.

[0013] The total converter loss is divided into two parts: series conduction loss and parallel switching loss. These are equivalent to series and parallel variable impedances connected to the mutual admittance interaction port, respectively. In each iteration of the subsequent power flow calculation, the real-time loss is first calculated based on the current transmission power and operating parameters. The two sets of variable impedance values ​​are updated synchronously and embedded into a unified impedance network model. Then, the actual transmission power deviation is inferred from the updated impedance, and the power reference value of the converter is corrected in reverse.

[0014] The loss calculation is only related to the amplitude of the transmitted power and is not affected by the power flow direction. The same loss calculation logic is used under rectification and inverter conditions. The access topology of the two types of variable impedance does not change with the switching of operating conditions. It is adapted to bidirectional power transmission only by adjusting the direction of the power reference value.

[0015] Furthermore, in the constraint system construction phase, a three-level absorption constraint set including electrical safety, protection action, and power quality is constructed. A hierarchical progressive trigger verification logic is adopted, with priority given to verifying the first-level electrical safety constraint. If the limit is exceeded, it is directly determined that the absorption boundary has been reached. If the previous level constraint is qualified, the next level constraint verification will proceed. The first level of electrical safety constraints includes AC node voltage deviation, branch current carrying capacity, and DC bus voltage deviation. The AC node voltage deviation threshold is dynamically adjusted according to the branch load rate. When the load rate is higher than 80%, the voltage deviation limit is tightened, and when the load rate is lower than 30%, the voltage deviation limit is relaxed. The second-level protection action constraints include the residual current protection action threshold, branch overcurrent protection setting value, and converter overload protection limit value. The residual current protection sets the warning threshold at 80% of the rated operating current and the over-limit boundary at 100%. The threshold is adjusted synchronously with the ambient temperature and humidity and the service life of the line. In high temperature and high humidity scenarios and old line scenarios, the action threshold is reduced proportionally. The third-level power quality constraints include three-phase voltage imbalance and DC-side ripple voltage coefficient. The baseline limit for the ripple voltage coefficient is 5% of the rated voltage of the DC bus. The threshold is adjusted according to the DC bus power fluctuation rate, and the faster the fluctuation rate, the stricter the limit.

[0016] Furthermore, in the typical scenario generation stage, historical photovoltaic power output data and AC / DC load operation data of the target area are collected. The daily photovoltaic power output peak, output fluctuation coefficient, output ramp-up rate, daily load peak-valley difference, and load fluctuation rate are used as the core feature dimensions. Before clustering, all feature data are normalized. The silhouette coefficient is used to determine the optimal number of clusters. The K-means clustering algorithm is used to divide the data into four seasons and combine the three weather types of sunny, cloudy, and rainy to generate the photovoltaic power output and load power time series curves of the basic typical days. Based on the nominal impedance of the cable and the typical deviation range of the project, three sets of discrete samples of high impedance, nominal impedance and low impedance are extracted. The three sets of parameter samples are coupled with the basic scenario to form a typical source-load and parameter coupling scenario. Together, they constitute a typical source-load and parameter coupling scenario set. Each time series curve corresponds to a time series coupling scenario, which includes a time section with a 15-minute interval. It matches the proportional characteristics of AC and DC loads under the corresponding season and covers the full operating scenarios of the transformer area and the discrete range of line parameters.

[0017] Furthermore, in the initial boundary solution stage, for each time segment of each time-series coupling scenario in the typical scenario set of source-load and parameter coupling, the total photovoltaic output is used as the incremental variable, and a variable step size incremental strategy is adopted. In the initial stage, a step size of 5% of the rated capacity of the transformer area is used for rapid approximation. When any warning threshold is triggered, the step size is switched to 0.5% of the rated capacity for search. The photovoltaic output is allocated to the corresponding DC node according to the capacity ratio of each grid connection point. The AC / DC alternating iterative power flow algorithm is used for calculation. First, the current photovoltaic output is substituted to calculate the DC subgrid power flow, and the DC side power of the converter and the voltage of each node are obtained. The values ​​of series variable impedance and parallel variable impedance and the corresponding losses are updated simultaneously. After completing the reverse correction of the power deviation, the power flow distribution of the AC subnet is solved based on the unified impedance network model to obtain the voltage of each node and the branch current on the AC side. The iterative process supports the switching between the converter rectification and inverter operating conditions, and the alternating iteration is repeated until the power flow on both sides converges. Determine whether the current operating condition triggers any constraint in the three-level power consumption constraint set. When the constraint is triggered for the first time, record the current total photovoltaic output, which is the initial power consumption boundary value of that time segment. After traversing all time segments, obtain the initial power consumption boundary time series curve for each scenario.

[0018] Furthermore, in the confidence boundary correction stage, impedance parameter sampling data of cables of the same type and service life are collected. At the same time, the resistance increase caused by conductor oxidation and the additional impedance caused by the contact resistance of line joints are considered. The impedance parameter probability distribution function of AC and DC lines in the transformer area is fitted. The impedance parameter follows a normal distribution, the mean is the nominal impedance value of the cable, and the standard deviation is obtained by statistical analysis of the sampling samples. Normal probability density function of line impedance:

[0019] The probability density function representing the line impedance parameters; This represents a random variable of line impedance, including the cable body impedance and the additional impedance of the line joints. The mathematical expectation (mean) of the line impedance is represented by the nominal impedance of the cable, plus the linear aging additional impedance caused by conductor oxidation. The standard deviation of line impedance includes parameter dispersion caused by cable manufacturing process and random discrete components of line joint contact resistance. It is obtained by statistical analysis of sampled data of cables of the same type and service life.

[0020] A probabilistic assessment model coupling the discreteness of line parameters and the temporal volatility of source load is constructed. The Monte Carlo sampling method is used to perform probabilistic calculations on the initial absorption boundary values ​​of each time section. The calculation process quantifies the coupling effect of the two types of uncertainties: the discreteness of parameters and the temporal volatility of source load. The system calculates the safe operation probability corresponding to different absorption capacities, extracts the absorption boundary intervals at three confidence levels (95%, 90%, and 85%), and fits a continuous mapping function between confidence level and absorption capacity. It can retrieve boundary values ​​at any confidence level as needed, simultaneously calculate the proportion of over-limit trigger probability of each constraint at different confidence levels, and output the probability contribution ranking of absorption bottleneck constraints.

[0021] Furthermore, in the boundary verification output stage, a simulation model of the transformer area protection device is built to restore the actual operating parameters of the protection device. The protection action logic adopts a stepped delay judgment to distinguish between two types of protection characteristics: instantaneous action and delayed action. Substitute the absorption boundary intervals at each confidence level into the simulation model of the transformer area protection device to perform steady-state over-limit verification and timing linkage verification. The steady-state over-limit verification simulates the operation logic of residual current protection, overcurrent protection, and undervoltage protection when the photovoltaic output is stable at the boundary value. The timing linkage verification simulates the transient process of the photovoltaic output rising to the boundary value at a typical ramp rate. By comparing the output ramp rate with the protection action delay, the transient over-limit risk is determined. For boundary values ​​that trigger protection malfunctions or pose transient risks, corresponding margin corrections are applied. The boundary values ​​corresponding to instantaneous action protection are corrected downwards by 3%, and the boundary values ​​corresponding to delayed action protection are corrected downwards by 8%. The corrected boundary values ​​are fed back to the unified impedance network model to update the boundary judgment parameters corresponding to the protection constraints. The final results are output after iterative updates, including photovoltaic absorption boundary curves for different seasons, weather conditions, and confidence levels, as well as the location results of the bottleneck nodes and bottleneck constraint types in the distribution area.

[0022] This invention also provides a photovoltaic absorption boundary assessment system for low-voltage AC / DC hybrid distribution areas, based on the above method, including a distribution area unified modeling module, a converter loss embedding module, an absorption constraint configuration module, a typical scenario generation module, an initial boundary solution module, a confidence boundary correction module, and a protection verification output module; The unified modeling module of the transformer area performs port equivalence and hierarchical division on the AC and DC side electrical nodes of the target transformer area, completes the Thevenin aggregation order reduction of the terminal nodes of the same branch, embeds the source load operation characteristics into the node self-admittance term in the form of admittance increment, configures the phase shift correction coupling term of the bidirectional converter, and constructs a unified impedance network model covering all AC and DC nodes. The converter loss embedding module collects converter conversion efficiency data under multiple operating conditions, fits the efficiency mapping function of multi-physical quantity coupling, and decomposes the total converter loss into two types of variable impedances connected in series and parallel to the mutual admittance interaction port. The impedance is updated synchronously in the power flow iteration and the power calculation deviation is corrected in reverse. The absorption constraint configuration module constructs a three-level absorption constraint set, configures hierarchical trigger verification logic and operating condition adjustment threshold, and sets a Boolean over-limit flag as the absorption boundary trigger criterion. The typical scenario generation module performs feature clustering based on historical operation data of the transformer area, couples discrete samples of line impedance, and generates a set of typical scenarios of source-load and parameter coupling. The initial boundary solution module uses a variable step size increment strategy and an alternating AC / DC iterative power flow algorithm to solve for the initial absorption boundary values ​​at each time section. The confidence boundary correction module fits the line impedance probability distribution and calculates the absorption boundary interval for multiple confidence levels through probability sampling. The protection verification output module restores the operating characteristics of the protection device, performs steady-state and time-series linkage verification, corrects the boundary values, and iteratively updates the output to show the final absorption boundary curve and bottleneck location results. The beneficial effects of this invention are as follows: 1. This invention performs port equivalence and hierarchical division on all AC and DC nodes, combined with Thevenin aggregation and order reduction processing of homogeneous terminal nodes, and converts the bidirectional converter into an equivalent mutual admittance interaction port with an adaptive phase shift correction coefficient. It constructs a unified impedance network model covering all nodes, restores the voltage coupling characteristics of DC-side photovoltaic power output fluctuations transmitted to the AC side through the converter, and eliminates the systematic bias caused by decoupling calculations. By constructing a converter efficiency mapping function through multi-condition sampling and quadratic surface fitting, the total loss is decomposed into series conduction loss and parallel switching loss and converted into a variable impedance embedded in the power flow iteration. In each round of calculation, the impedance parameters are updated synchronously and the power deviation is corrected in reverse. It dynamically adapts to the modulation effect of transmission power ratio, ambient temperature, and DC bus voltage on conversion efficiency, and reduces the reverse evaluation error during photovoltaic peak and off-peak periods.

[0023] 2. This invention constructs a three-level progressive set of absorption constraints for electrical safety, protection action, and power quality. It adopts a hierarchical trigger verification logic. The electrical safety constraint threshold is dynamically adjusted according to the branch load rate. The protection action constraint is incorporated into the protection settings for residual current, overcurrent, and converter overload. The action threshold for high-temperature and high-humidity lines and old lines is dynamically lowered in combination with the ambient temperature and humidity and the service life of the line. The power quality constraint sets a limit for DC ripple voltage that is dynamically tightened with the power fluctuation rate. A simulation model of the protection device is built to distinguish between instantaneous and delayed action protection characteristics. Dual verification of steady-state over-limit and time-sequence linkage is performed, and corresponding margin corrections are applied according to the protection type. This avoids hidden risks such as protection maloperation and power quality exceeding standards in actual operation, and avoids the waste of absorption space caused by uniform conservative settings.

[0024] 3. This invention uses multi-dimensional feature clustering based on historical operating data of the transformer substation to generate typical source-load time-series scenarios covering all four seasons and three types of weather. It also couples high, nominal, and low set of discrete line impedance samples to form a scenario set covering the entire operating range. By fitting the probability distribution of line impedance considering conductor oxidation and joint contact resistance, a probability assessment model coupling the discreteness of line parameters and the temporal volatility of source-load time-series is constructed. Monte Carlo probability sampling is used to calculate and output the absorption boundary intervals and continuous mapping functions at multiple confidence levels, allowing for flexible selection of corresponding boundary values ​​based on the risk tolerance level of the transformer substation. Simultaneously, the invention statistically analyzes the probability of exceeding limits for each constraint at different confidence levels, locating the absorption bottleneck nodes and bottleneck constraint types, providing a tiered and flexible decision-making basis for photovoltaic grid connection in the transformer substation. Attached Figure Description

[0025] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a flowchart of the initial boundary solution process of this invention. Detailed Implementation

[0026] 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 embodiments of the present invention, and not all embodiments. 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.

[0027] like Figures 1 to 2 As shown, this embodiment of the invention provides a photovoltaic absorption boundary assessment method for low-voltage AC / DC hybrid distribution areas, including a unified impedance modeling stage, a loss equivalent embedding stage, a constraint system construction stage, a typical scenario generation stage, an initial boundary solution stage, a confidence boundary correction stage, and a boundary verification output stage. In this embodiment of the invention, during the unified impedance modeling stage, the AC distribution cables, DC busbars, bidirectional converters, photovoltaic grid-connected nodes, and AC / DC load nodes of the target transformer area are port equivalent. On the AC side, nodes are divided step by step according to the low-voltage outgoing terminals of the distribution transformer, branch box nodes, and end user nodes. On the DC side, nodes are divided according to the photovoltaic combiner port, energy storage access port, and DC load port. The node numbering adopts continuous integer encoding, and AC nodes and DC nodes are numbered in separate intervals to avoid coupling term index conflicts; Thevenin equivalent aggregation is performed on multiple end nodes with the same electrical characteristics under the same distribution branch, and the end user node is equivalent to a single aggregation port, thereby compressing the admittance matrix dimension; The criteria for determining electrical characteristics include three indicators: the rated voltage level of the node, the type and length of the access cable, and the rated capacity range of the single household load. End nodes whose deviations in all three indicators are controlled within 10% are included in the same aggregation group. The equivalent impedance of the port after aggregation is obtained by equivalent conversion of the branch trunk impedance and the impedance of all parallel user branches. The total power of the aggregation port is the algebraic sum of the power of all nodes in the group. The aggregation node retains the original branch electrical topology connection relationship and voltage level attributes, and does not change the impedance structure of the main network of the transformer area.

[0028] The constant power external characteristics of the photovoltaic grid-connected unit, the droop control characteristics of the energy storage unit, and the impedance characteristics of the DC load are embedded into the self-admittance terms of the corresponding nodes in the form of admittance increments. At the same time, an adaptive phase shift correction coefficient is set for the mutual admittance coupling terms of the bidirectional converter to adapt to the phase shift deviation of the converter caused by voltage distortion under low-voltage weak grid conditions. The aggregated AC subgrid node admittance matrix and DC subgrid node admittance matrix are constructed respectively. The adaptive phase shift correction coefficient uses the total harmonic distortion rate of the AC side grid connection point voltage and the real-time transmission power factor of the converter as input parameters. The correction mapping relationship is obtained by fitting the measured phase shift deviation data under typical operating conditions of the low-voltage distribution area. The harmonic distortion rate is collected from the voltage sampling value of the AC side grid connection point of the converter, and the power factor is calculated in real time from the active and reactive power currently transmitted.

[0029] When the harmonic distortion rate increases and the power factor decreases, the phase shift correction amplitude should be increased accordingly. The correction range is limited to within ±15% of the rated phase shift to avoid overcorrection that could cause instability in the coupling terms. This coefficient is updated synchronously with the current operating parameters in each power flow iteration calculation and directly affects the phase angle component of the mutual admittance coupling terms.

[0030] The DC load uses constant admittance parameters to be directly superimposed on the node self-admittance; the energy storage unit converts the voltage-power regulation characteristics of droop control into admittance increments that dynamically change with the node voltage; the constant power characteristics of the photovoltaic grid-connected unit calculate the real-time equivalent admittance and update the self-admittance value in each power flow iteration based on the current node voltage and target output.

[0031] The bidirectional converter is equivalent to the mutual admittance interaction port connecting the AC and DC subnets. Combining the power transmission characteristics of the converter in both rectification and inversion modes, the converter port voltage ratio and the corrected phase shift characteristics are transformed into coupling terms in the admittance matrix, ultimately forming a unified impedance network model that includes all AC and DC nodes.

[0032] The coupling terms in the model can automatically adapt to the bidirectional operation of the converter. Under inverter operation, power flows from the DC side to the AC side, and the coupling terms correspond to the admittance relationship in the power delivery direction. Under rectifier operation, power flows from the AC side to the DC side, and the coupling terms can match the reverse power transfer characteristics by simply adjusting the sign of the components.

[0033] In this embodiment of the invention, during the loss equivalent embedding stage, the conversion efficiency data of the converter under different transmission power ratios, ambient temperatures, and DC bus voltages are collected through bench testing or simulation. The transmission power ratio is sampled at 5% intervals in the 0-10% light load range and traversed at 10% intervals in the 10% to rated power range. The ambient temperature covers the typical operating range of the transformer area from -10℃ to 45℃. The least squares method is used to fit the quadratic surface. The fitting process takes the minimum mean square error as the objective function. For outliers, a weighted regression method is used to reduce the corresponding weights, thereby weakening the interference of outlier data on the fitting results. The coupling functions of converter efficiency, power, temperature and voltage are obtained. The sampling point weights are allocated based on the actual probability of the corresponding operating conditions occurring in the transformer area throughout the year. Sampling points in commonly used operating ranges near the rated load are given the highest baseline weights, while the weights of sampling points in low-probability operating conditions such as light loads and extreme temperatures are proportionally reduced to ensure the fitting accuracy of commonly used operating ranges.

[0034] Sampling points whose fitting residuals exceed 3 times the mean square error are identified as outliers, and their weights are reduced to one-fifth of the baseline weights. The fitting adopts an iterative reweighting method. After each round of fitting, the residuals of all sampling points are recalculated and the weights are updated synchronously. The process terminates after a maximum of 5 rounds of iteration.

[0035] The total converter loss is divided into two parts: series conduction loss and parallel switching loss. These are equivalent to series and parallel variable impedances connected to the mutual admittance interaction port, respectively. In each iteration of the subsequent power flow calculation, the real-time loss is first calculated based on the current transmission power and operating parameters. The two sets of variable impedance values ​​are updated synchronously and embedded into a unified impedance network model. Then, the actual transmission power deviation is inferred from the updated impedance, and the power reference value of the converter is corrected in reverse to eliminate the power flow calculation deviation caused by the loss.

[0036] The ratio of the two types of losses is determined based on the device loss test data under the rated operating conditions of the converter. The conduction loss includes the on-state voltage drop loss of the power device and the freewheeling diode loss, and its proportion increases linearly with the amplitude of the transmission current. The switching loss includes the device turn-on and turn-off losses and the buffer circuit loss, and its proportion is adjusted with the fluctuation of DC bus voltage and switching frequency.

[0037] The series variable impedance is calculated from the current port current RMS value and the conduction loss power, and is connected in series to the equivalent branch of the main power circuit of the converter; the parallel variable impedance is calculated from the current port voltage RMS value and the switching loss power, and is connected in parallel between the port and the reference node. After each iteration completes the loss calculation, the values ​​of the two types of impedances are updated synchronously.

[0038] The corrected power reference value is applied to the iterative calculation of both AC and DC subnets. The power correction direction is adaptively adjusted according to the converter operating conditions: under inverter conditions, the AC output power is obtained by subtracting losses from the DC input power, and under rectifier conditions, the DC output power is obtained by subtracting losses from the AC input power.

[0039] The correction process only updates the power boundary value at the converter port and does not change the injected power setting value at the source-load node, ensuring that the calculation benchmark of the photovoltaic output increment variable remains consistent; the power correction is performed only once in each round of alternating iterations.

[0040] In this embodiment of the invention, during the constraint system construction stage, a three-level set of absorption constraints including electrical safety, protection action, and power quality is constructed. A hierarchical progressive trigger verification logic is adopted, prioritizing the verification of the first-level electrical safety constraint. If the limit is exceeded, it is directly determined that the absorption boundary has been reached. If the previous level constraint is qualified, the next level constraint verification will proceed. The single-round verification only starts the verification of the next level of constraints if all previous levels of constraints have not exceeded the limit. If any level of constraint triggers an exceedance, the entire verification process will be terminated immediately, and the current working condition will be marked as having reached the absorption boundary.

[0041] Verification priorities are ranked according to the severity of the fault impact. Electrical safety faults pose the highest risk of causing equipment damage and power outages, therefore they are set as the highest verification level. Warning thresholds are uniformly set at 90% of the corresponding over-limit threshold. Step size switching is triggered only during the large step size phase of variable step size search; warning signals are no longer responded to during the small step size fine search phase.

[0042] The first level of electrical safety constraints includes AC node voltage deviation, branch current carrying capacity, and DC bus voltage deviation. The AC node voltage deviation threshold is dynamically adjusted according to the branch load rate. When the load rate is higher than 80%, the voltage deviation limit is tightened, and when the load rate is lower than 30%, the voltage deviation limit is relaxed. The voltage deviation reference limit is set at ±7% of the rated voltage, and the branch load rate is the ratio of the current transmission power of the branch to the power corresponding to the rated current of the branch.

[0043] When the load rate is higher than 80%, the line voltage drop ratio increases significantly, the voltage operating margin narrows, and the voltage deviation limit is tightened to ±5% of the rated voltage. When the load rate is lower than 30%, the line voltage drop has a negligible impact, the voltage adjustment space is sufficient, and the voltage deviation limit is relaxed to ±10% of the rated voltage. When the load rate is between 30% and 80%, the baseline limit remains unchanged, and the threshold of each branch is calculated and adjusted independently according to its own load rate.

[0044] The second-level protection action constraints include the residual current protection action threshold, branch overcurrent protection setting value, and converter overload protection limit value. The residual current protection sets the warning threshold at 80% of the rated operating current and the over-limit boundary at 100%. The threshold is adjusted synchronously with the ambient temperature and humidity and the service life of the line. In high temperature and high humidity scenarios and old line scenarios, the action threshold is reduced proportionally. The reference operating temperature is 25℃, and the reference relative humidity is 60%. For every 10℃ increase in ambient temperature beyond the reference operating temperature, the corresponding residual current operating threshold is reduced by 3%; for every 20% increase in ambient relative humidity beyond the reference humidity, the corresponding operating threshold is reduced by 2%; and for every 20% increase in the line's service life beyond its design life, the corresponding overcurrent protection setting value is reduced by 5%.

[0045] The various correction coefficients are added together by multiplication, and the cumulative reduction does not exceed 25% of the rated threshold to avoid frequent false triggering of protection due to excessive tightening of the threshold; the correction process retains the individual records of each factor, so that the influence of a single factor on the threshold can be traced separately.

[0046] The third-level power quality constraints include three-phase voltage imbalance and DC-side ripple voltage coefficient. The baseline limit for the ripple voltage coefficient is 5% of the rated voltage of the DC bus. The threshold is adjusted according to the DC bus power fluctuation rate, and the faster the fluctuation rate, the stricter the limit.

[0047] The power fluctuation rate is calculated as the ratio of the change in DC bus power per unit time to the rated power. When the fluctuation rate is less than 5% of the rated power per minute, the baseline limit remains unchanged; when the fluctuation rate is between 5% and 15% of the rated power per minute, the limit is tightened to 80% of the baseline value; when the fluctuation rate is greater than 15% of the rated power per minute, the limit is tightened to 60% of the baseline value; the limit for each time segment is determined in real time based on the power fluctuation rate of the current time period.

[0048] In this embodiment of the invention, during the typical scenario generation stage, historical photovoltaic power output data and AC / DC load operation data of the target area are collected. The daily photovoltaic power output peak value, power output fluctuation coefficient, power output ramp-up rate, daily load peak-valley difference, and load fluctuation rate are used as the core feature dimensions. Before clustering, all feature data are normalized. The silhouette coefficient is used to determine the optimal number of clusters. The K-means clustering algorithm is used to divide the data into four seasons (spring, summer, autumn, and winter) and combined with three weather types (sunny, cloudy, and rainy) to generate the photovoltaic power output and load power time series curves for the basic typical days. The input to the K-means clustering algorithm is a standardized feature vector split by natural day. Each feature vector corresponds to the complete running data of a day. Each feature dimension is mapped to the interval of 0 to 1 after range normalization. The clustering calculation uses Euclidean distance as the sample similarity metric and iteratively updates the center coordinates of each cluster until the cluster center offset is less than the set convergence threshold or the preset maximum number of iterations is reached. After clustering is completed, the central feature vectors of each cluster are denormalized to restore the corresponding time series curves. Then, the weather labels in the historical data are matched to classify and label the clustering results, and finally, typical daily photovoltaic output and load power curves corresponding to the season and weather type are formed.

[0049] Based on the nominal impedance of the cable and the typical deviation range of the project, three sets of discrete samples of high impedance, nominal impedance and low impedance are extracted. The three sets of parameter samples are coupled with the basic scenario to form a typical source-load and parameter coupling scenario. Together, they constitute a typical source-load and parameter coupling scenario set. Each time series curve corresponds to a time series coupling scenario, which includes a time section with a 15-minute interval. It matches the proportional characteristics of AC and DC loads under the corresponding season and covers the full operating scenarios of the transformer area and the discrete range of line parameters.

[0050] Each set of basic typical daily scenarios is paired one-to-one with three sets of parameter samples: high impedance, nominal impedance, and low impedance. Each pairing result corresponds to an independent coupling scenario. Among them, the high impedance sample corresponds to the extreme operating conditions of old cables and long-distance branches, the nominal impedance sample corresponds to the design benchmark operating conditions of the transformer area, and the low impedance sample corresponds to the optimal operating conditions of newly commissioned short-distance cables.

[0051] All time segments within the same scenario share the same set of line impedance parameters. The source-load time series data changes dynamically with the time segment. The coupled scenario set is stored in three categories: season, weather, and impedance level. Each scenario contains complete intraday data for 96 time segments.

[0052] The clustering process simultaneously extracts the feature dimensions of AC and DC loads and performs joint clustering according to the same time dimension to ensure that the time-series proportion of AC and DC loads in the generated typical daily curves is consistent with the actual operating statistics of the corresponding season; the proportion of DC load fluctuates synchronously with seasonal temperature changes, with the proportion of cooling DC load increasing in summer and the proportion of heating DC load adjusting accordingly in winter.

[0053] In this embodiment of the invention, during the initial boundary solution stage, for each time segment of each time-series coupling scenario in the typical scenario set of source-load and parameter coupling, the total photovoltaic output is used as an increasing variable, and a variable step-size increasing strategy is adopted. In the initial stage, a step size of 5% of the rated capacity of the transformer area is used for rapid approximation. When any warning threshold is triggered, the step size is switched to 0.5% of the rated capacity for searching. The photovoltaic output is allocated to the corresponding DC node according to the capacity ratio of each grid connection point. When switching step sizes, the output increment is simultaneously rolled back by a large step size. Starting from the initial value of a small step size, the search is performed by gradually increasing the increment to avoid initial value deviation caused by large step sizes directly crossing the boundary interval. During the search process, the output value that first triggers the limit constraint and the output value that did not trigger the limit constraint in the previous step size are recorded simultaneously. The median value of the two sets of output values ​​is used for supplementary verification. Finally, the minimum output value that first triggers the limit constraint is used as the initial absorption boundary of the time section.

[0054] The AC / DC alternating iterative power flow algorithm is used for calculation. First, the current photovoltaic output is substituted to calculate the DC subgrid power flow, and the DC side power of the converter and the voltage of each node are obtained. The values ​​of series variable impedance and parallel variable impedance and the corresponding losses are updated simultaneously. After completing the reverse correction of the power deviation, the power flow distribution of the AC subnet is solved based on the unified impedance network model to obtain the node voltage and branch current on the AC side. The iterative process supports switching between converter rectification and inverter operation. The iterative process is repeated until the power flow on both sides converges. The convergence criterion is set to the node voltage amplitude deviation between two adjacent iterations being less than 10. -4 pu (per unit value); The converter operating condition is automatically determined by the net power flow direction on the DC side. When the total output of photovoltaic and energy storage on the DC side is greater than the total power of the DC load, the converter operates in inverter mode, and power flows from the DC subgrid to the AC subgrid; otherwise, it operates in rectification mode. Each iterative cycle first completes the DC subgrid power flow calculation, updates the converter port power and impedance parameters, and then solves for the AC subgrid power flow. The convergence criterion includes both node voltage amplitude deviation and branch power deviation. The voltage deviation threshold and power deviation threshold are matched to the calculation accuracy requirements of the low-voltage distribution area. If one side converges first, the parameters of that side remain unchanged, and the other side continues to iterate until both sides meet the convergence condition at the same time.

[0055] Determine whether the current operating condition triggers any constraint in the three-level power consumption constraint set. When the constraint is triggered for the first time, record the current total photovoltaic output, which is the initial power consumption boundary value of that time segment. After traversing all time segments, obtain the initial power consumption boundary time series curve for each scenario.

[0056] Electrical safety constraints and protection action constraints adopt a single-node triggering judgment rule. If any node or branch in the distribution area triggers the corresponding threshold, it is judged as an overall over-limit, giving priority to ensuring the safety of power grid operation. Power quality constraints adopt a global worst value judgment rule, taking the node with the most serious index of the entire network for verification. A single node over-limit means that the power quality does not meet the operating requirements under the current output.

[0057] In this embodiment of the invention, during the confidence boundary correction stage, impedance parameter sampling data of cables of the same type and service life are collected. At the same time, the resistance increase caused by conductor oxidation and the additional impedance caused by the contact resistance of line joints are considered. The impedance parameter probability distribution function of AC and DC lines in the transformer area is fitted. The impedance parameter follows a normal distribution, the mean is the nominal impedance value of the cable, and the standard deviation is obtained by statistical analysis of the sampling samples. The additional impedance caused by conductor oxidation increases linearly with the service life of the line. The incremental ratio of the nominal impedance is converted according to the actual service life and included in the mean correction term of the probability distribution. The contact resistance of the line joints exhibits discrete random distribution characteristics and is included as an additional random component in the standard deviation calculation of the probability distribution. The final fitted distribution simultaneously covers the impedance increment caused by the deviation of the cable body manufacturing parameters and the aging during operation.

[0058] A probabilistic assessment model coupling the discreteness of line parameters and the temporal volatility of source load is constructed. The Monte Carlo sampling method is adopted, and 1000 sampling times are set. The initial absorption boundary value of each time section is probabilistically calculated. The calculation process quantifies the coupling effect of the two types of uncertainties: parameter discreteness and temporal volatility of source load. The sampling process adopts a stratified joint sampling method. The first layer is the source-load time sequence scenario layer, and the sampling weight is allocated according to the historical occurrence probability of each typical scenario. The second layer is the line parameter layer, where the impedance of each branch is sampled independently and the parameter correlation of the same batch of cables is retained.

[0059] First, the source load operating parameters for the corresponding time segment are extracted. Then, impedance parameter samples of the corresponding branches are randomly selected from the line impedance probability distribution. The two sets of parameters are simultaneously substituted into the power flow calculation model to obtain the single-sample absorption boundary value. The coefficient of variation threshold for sampling termination is set to 3%, and sampling stops when the threshold is reached. During the sampling process, the constraint type corresponding to each limit exceedance is recorded simultaneously.

[0060] The discreteness of the default line parameters and the temporal volatility of the source load are calculated as independent random variables. The two types of factors are sampled independently and then combined and substituted into the power flow calculation model. The coupling effect is reflected in the fact that the fluctuation amplitude of the absorption boundary under the combined action of the two factors is significantly greater than the fluctuation amplitude under the action of a single factor alone. The dispersion of the boundary value distribution is higher, which can more realistically reflect the comprehensive uncertainty risk in the actual operation of the transformer area.

[0061] The system calculates the safe operation probability corresponding to different absorption capacities, extracts the absorption boundary intervals at three confidence levels (95%, 90%, and 85%), and fits a continuous mapping function between confidence level and absorption capacity. It can retrieve boundary values ​​at any confidence level as needed, simultaneously calculate the proportion of over-limit trigger probability of each constraint at different confidence levels, and output the probability contribution ranking of absorption bottleneck constraints.

[0062] The constraint contribution is calculated as the proportion of the number of samplings that trigger limit exceedances by a single type of constraint to the total number of samplings for all limit exceedances. The constraint with the highest proportion is determined as the primary bottleneck constraint under the current confidence level. The electrical node that first triggers the limit value in each limit exceedance event is recorded simultaneously and sorted from high to low according to the frequency of node limit exceedances. The node with the highest frequency is determined as the bottleneck node of the distribution area. The location result can be directly mapped to the specific power distribution branch and source load access port of the distribution area.

[0063] In this embodiment of the invention, during the boundary verification output stage, a simulation model of the transformer area protection device is built to restore the actual operating parameters of the protection device. The protection action logic adopts a stepped delay judgment to distinguish between two types of protection characteristics: instantaneous action and delayed action. Residual current protection and converter overload protection correspond to instantaneous action settings. The action delay is set according to the inherent action time of the protection device, without adding additional delay margin. The protection action is triggered immediately after the limit is exceeded. Branch overcurrent protection corresponds to stepped delay settings, with three action levels, corresponding to 1.2 times, 1.5 times, and 2 times the rated setting value, respectively. The higher the limit exceedance multiple, the shorter the corresponding action delay. During the verification process, the corresponding delay setting is first matched according to the limit exceedance magnitude of the electrical quantity, and then the duration of the limit exceedance is compared with the corresponding action delay to accurately determine whether the protection has triggered an action.

[0064] Substitute the absorption boundary intervals at each confidence level into the simulation model of the transformer area protection device to perform steady-state over-limit verification and timing linkage verification. The steady-state over-limit verification simulates the operation logic of residual current protection, overcurrent protection, and undervoltage protection when the photovoltaic output is stable at the boundary value. The timing linkage verification simulates the transient process of the photovoltaic output rising to the boundary value at a typical ramp rate. By comparing the output ramp rate with the protection action delay, the transient over-limit risk is determined. The verification process sets three typical photovoltaic ramp rates, corresponding to three typical scenarios: stable ramp rate on sunny days, slow fluctuation on cloudy days, and rapid fluctuation on partly cloudy days. The simulation time step is set to 10 milliseconds to match the normal sampling period of the low-voltage protection device.

[0065] Electrical quantity values ​​are collected step-by-step and matched with the stepped delay judgment logic of the protection device. The protection delay is divided into three levels: Stage I instantaneous action, Stage II short delay, and Stage III long delay. If the duration of the electrical quantity exceeding the limit exceeds the corresponding protection's action delay threshold, a transient action risk is determined. The shorter the duration of the exceeding limit, the lower the corresponding transient risk level; conversely, a larger boundary margin correction needs to be applied.

[0066] For boundary values ​​that trigger protection malfunctions or pose transient risks, corresponding margin corrections are applied. The boundary values ​​corresponding to instantaneous action protection are corrected downwards by 3%, and the boundary values ​​corresponding to delayed action protection are corrected downwards by 8%. The corrected boundary values ​​are fed back to the unified impedance network model to update the boundary judgment parameters corresponding to the protection constraints. The final results are output after iterative updates, including photovoltaic absorption boundary curves for different seasons, weather conditions, and confidence levels, as well as the location results of the bottleneck nodes and bottleneck constraint types in the distribution area.

[0067] The iteration process uses the relative deviation of the boundary values ​​after two consecutive corrections being less than 1% as the convergence criterion. At the same time, it is verified that the corrected boundary values, when substituted into the protection simulation model, no longer trigger protection actions. The iteration is considered to have converged when both conditions are met. If the convergence state is not reached after three consecutive iterations, the minimum boundary value after each correction is taken as the final result, prioritizing the safety of the transformer area operation.

[0068] If the same boundary value triggers multiple types of protection actions simultaneously, the correction coefficients are calculated by multiplication, with priority given to the protection with higher action level and shorter delay as the main correction basis; if multiple branches or nodes under the same type of protection trigger limit exceedances simultaneously, the correction range corresponding to the most stringent node is applied, and the correction coefficients of multiple nodes are not superimposed repeatedly.

[0069] This invention also provides a photovoltaic absorption boundary assessment system for low-voltage AC / DC hybrid distribution areas, based on the above method, including a distribution area unified modeling module, a converter loss embedding module, an absorption constraint configuration module, a typical scenario generation module, an initial boundary solution module, a confidence boundary correction module, and a protection verification output module; The unified modeling module of the transformer area performs port equivalence and hierarchical division on the AC and DC side electrical nodes of the target transformer area, completes the Thevenin aggregation order reduction of the terminal nodes of the same branch, embeds the source load operation characteristics into the node self-admittance term in the form of admittance increment, configures the phase shift correction coupling term of the bidirectional converter, and constructs a unified impedance network model covering all AC and DC nodes. The converter loss embedding module collects converter conversion efficiency data under multiple operating conditions, fits the efficiency mapping function of multi-physical quantity coupling, and decomposes the total converter loss into two types of variable impedances connected in series and parallel to the interactive port. In the power flow iteration, the impedance is updated synchronously and the power calculation deviation is corrected in reverse. The absorption constraint configuration module constructs a three-level absorption constraint set, configures hierarchical trigger verification logic and operating condition adjustment threshold, and sets a Boolean over-limit flag as the absorption boundary trigger criterion. The typical scenario generation module performs feature clustering based on historical operation data of the transformer area, couples discrete samples of line impedance, and generates a set of typical scenarios of source-load and parameter coupling. The initial boundary solution module uses a variable step size increment strategy and an alternating AC / DC iterative power flow algorithm to solve for the initial absorption boundary values ​​at each time section. The confidence boundary correction module fits the line impedance probability distribution and obtains the absorption boundary interval of multiple confidence levels through probability sampling calculation. The protection verification output module restores the operating characteristics of the protection device, performs steady-state and time-series linkage verification, corrects the boundary values, and iteratively updates the output of the final absorption boundary curve and bottleneck location results.

[0070] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0071] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for assessing the photovoltaic absorption boundary of a low-voltage AC / DC hybrid distribution area, characterized in that, The specific steps include the following: The electrical nodes of the low-voltage AC / DC hybrid distribution area are divided into port equivalents and hierarchical divisions. The node admittance matrices of the AC and DC subnets are constructed respectively. The bidirectional converter is equivalent to the mutual admittance interaction port. Combined with the AC / DC coupling characteristics of the bidirectional converter, a unified impedance network model is constructed. The conversion efficiency data of the bidirectional converter is collected and the efficiency mapping function is fitted. The converter loss is equivalent to a variable impedance connected to the mutual admittance interaction port. The impedance is updated synchronously and the power reference value of the bidirectional converter is corrected in reverse during the power flow iteration process. A three-level set of absorption constraints, including electrical safety, protection actions, and power quality, is constructed. A hierarchical progressive trigger verification logic is adopted, with constraint over-limit triggering as the basis for determining the absorption boundary. Based on historical operation data of the transformer substation, feature clustering is performed, and typical scenario sets of source-load and parameter coupling are generated by combining operation parameters, covering the entire operating range of the transformer substation. For each time section in typical scenarios, a variable step size incremental strategy is adopted with the total photovoltaic output as the variable, and the convergence calculation is completed by the AC / DC alternating iterative power flow algorithm. The initial absorption boundary value is determined and recorded according to the constraint triggering conditions. The probability distribution function of the line impedance parameters is fitted, a probability assessment model coupled with the two dimensions of parameter discreteness and source load time-series fluctuation is constructed, the probability distribution of the absorption boundary is calculated, the absorption boundary interval under different confidence levels is extracted, and the absorption bottleneck constraint is located. A simulation model of the transformer area protection device was built and the protection action characteristics were restored. The steady-state over-limit and time-series linkage verification of the absorption boundary under each confidence level was carried out. The boundary value was corrected according to the protection type. After iteration, the absorption boundary curve and bottleneck location results were output.

2. The method for assessing the photovoltaic absorption boundary of a low-voltage AC / DC hybrid distribution area according to claim 1, characterized in that, When constructing the unified impedance network model, port equivalence is performed on the AC power distribution cables, DC bus, bidirectional converter, photovoltaic grid-connected node, and AC / DC load node of the target transformer area; On the AC side, electrical nodes are divided into levels according to the low-voltage outgoing terminals of the distribution transformer, branch box nodes, and end-user nodes. On the DC side, electrical nodes are divided into photovoltaic combiner ports, energy storage access ports, and DC load ports. Thevenin equivalent aggregation reduction is performed on the AC side end nodes with consistent electrical characteristics under the same distribution branch. The operating characteristics of DC side photovoltaic, energy storage and load are embedded into the self-admittance terms of the corresponding nodes in the form of admittance increments. A phase shift correction coupling term is set for the bidirectional converter, and finally the unified impedance network model containing all AC and DC nodes is formed.

3. The photovoltaic absorption boundary assessment method for a low-voltage AC / DC hybrid distribution area according to claim 1, characterized in that, When fitting the efficiency mapping function, the conversion efficiency data of the bidirectional converter under different transmission power ratios, ambient temperatures, and DC bus voltages are collected. The least squares method is used to perform quadratic surface fitting. The fitting process takes the minimum mean square error as the objective function. The weighted regression method is used to reduce the weights corresponding to outliers to obtain the efficiency mapping function of multi-physical quantity coupling. The converter loss is divided into two parts: series conduction loss and parallel switching loss, which are equivalent to series variable impedance and parallel variable impedance, respectively, and both are connected to the interaction port. In each round of calculation during the power flow iteration process, the real-time loss is first calculated based on the current transmission power and operating parameters. The values ​​of the two types of variable impedances are then updated synchronously and embedded into the unified impedance network model. Finally, the power reference value of the bidirectional converter is corrected in reverse by the updated impedance.

4. The method for assessing the photovoltaic absorption boundary of a low-voltage AC / DC hybrid distribution area according to claim 1, characterized in that, The hierarchical progressive triggering verification logic is used for the three-level absorption constraint set. The first-level electrical safety constraint is verified first. If the constraint corresponding to the current verification level exceeds the limit, it is directly determined that the photovoltaic output has reached the absorption boundary. The next level constraint verification is only entered when the previous level constraint is qualified. The first level of electrical safety constraints includes AC node voltage deviation, branch current carrying capacity, and DC bus voltage deviation. The second level of protection action constraints includes residual current protection action threshold, branch overcurrent protection setting value, and the overload protection limit value of the bidirectional converter. The third level of power quality constraints includes three-phase voltage imbalance and DC side ripple voltage coefficient.

5. The photovoltaic absorption boundary assessment method for a low-voltage AC / DC hybrid distribution area according to claim 4, characterized in that, The threshold values ​​for each level of constraint are dynamically adjusted according to the operating conditions: the AC node voltage deviation threshold value of the electrical safety constraint is dynamically adjusted according to the branch load rate, and the higher the load rate, the stricter the limit. The threshold for the protection action constraint is dynamically adjusted according to the ambient temperature and humidity and the service life of the line. In high temperature and high humidity scenarios and old line scenarios, the corresponding threshold is reduced proportionally. The limit of the DC side ripple voltage coefficient of the power quality constraint is dynamically adjusted according to the DC bus power fluctuation rate; the faster the fluctuation rate, the stricter the limit.

6. The method for assessing the photovoltaic absorption boundary of a low-voltage AC / DC hybrid distribution area according to claim 1, characterized in that, When generating the typical scenario set, historical photovoltaic power output data and AC / DC load operation data of the target area are collected. After normalizing the core dimension operation data, including photovoltaic power output characteristics and load operation characteristics, the K-means clustering algorithm is used to divide the data into four seasons and combine multiple weather types to generate the photovoltaic power output and load power time series curves of basic typical days. Multiple sets of discrete line impedance parameters are extracted and coupled one by one with the basic scenarios to form a typical scenario set of source-load-parameter coupling, covering the entire operating range of the transformer area and the discrete range of line parameters.

7. The photovoltaic absorption boundary assessment method for a low-voltage AC / DC hybrid distribution area according to claim 6, characterized in that, When solving the initial absorption boundary value, for each time section of each time series scenario, the variable step size incremental strategy is adopted with the total photovoltaic output as the incremental variable. In the initial stage, a large step size is used to quickly approach the boundary region. After approaching the constraint over-limit threshold, the small step size is switched to fine search. The photovoltaic output is allocated to the corresponding DC node according to the capacity ratio of each grid connection point. The AC / DC alternating iterative power flow algorithm is used to solve the power flow distribution of the DC subnet and the AC subnet in turn. The iterative process supports the switching between rectification and inverter modes of the bidirectional converter. The power flow of the two subnets is solved alternately until the power flow calculation converges. The total photovoltaic output recorded when the constraint is first triggered is the initial absorption boundary value of the corresponding time section.

8. The method for assessing the photovoltaic absorption boundary of a low-voltage AC / DC hybrid distribution area according to claim 7, characterized in that, When constructing the probability assessment model, the sampling data of the line impedance parameters are collected, and the effects of conductor oxidation and the additional impedance caused by the contact resistance of the line joints are considered. The probability distribution function of the impedance parameters of the AC / DC line in the transformer area is obtained by fitting. A probability sampling method is used to calculate the probability distribution of the absorption boundary of each time section, and to quantify the coupling effect of the two types of uncertainties: parameter discreteness and source load temporal volatility. The safe operation probability corresponding to different absorption capacities is statistically analyzed, the absorption boundary intervals under multiple confidence levels are extracted, the probability ratio of each constraint exceeding the limit is statistically analyzed, and the absorption bottleneck constraint results are output in order of probability contribution.

9. The photovoltaic absorption boundary assessment method for a low-voltage AC / DC hybrid distribution area according to claim 8, characterized in that, During boundary verification, a simulation model of the protection device in the transformer area is built to restore the actual operating parameters of the protection device. The protection action logic adopts a stepped delay judgment to distinguish between two types of protection action characteristics: instantaneous action and delayed action. Substitute the absorption boundary intervals at each confidence level into the simulation model, and perform steady-state over-limit verification and time-series linkage verification respectively. The steady-state over-limit verification is used to simulate the protection action behavior under the stable operating condition of photovoltaic power output, and the time-series linkage verification is used to assess the transient over-limit risk during the photovoltaic power output ramp-up process. For boundary values ​​that have the risk of steady-state or transient over-limit, the margin coefficient corresponding to the protection type is adjusted downward. The adjusted boundary values ​​are fed back to the unified impedance network model and the corresponding constraint parameters are updated. After iterative updates, the absorption boundary curves and bottleneck location results are output for different scenarios and confidence levels.

10. A photovoltaic absorption boundary assessment system for low-voltage AC / DC hybrid distribution areas, based on the method described in any one of claims 1-9, characterized in that, It includes a unified modeling module for transformer areas, a converter loss embedding module, a consumption constraint configuration module, a typical scenario generation module, an initial boundary solution module, a confidence boundary correction module, and a protection verification output module; The unified modeling module for the transformer substation is used to perform port equivalence and hierarchical division of the electrical nodes on the AC / DC side of the transformer substation, complete the aggregation and order reduction of the end nodes, and construct a unified impedance network model covering all AC / DC nodes. The converter loss embedding module is used to collect the conversion efficiency data of the bidirectional converter under multiple operating conditions, fit the efficiency mapping function, and convert the converter loss into a variable impedance connected to the interaction port. During the power flow iteration process, the impedance is updated synchronously and the power deviation is corrected in reverse. The absorption constraint configuration module is used to construct the three-level absorption constraint set, configure the hierarchical progressive trigger verification logic and the constraint threshold that is dynamically adjusted according to the working conditions, and set the over-limit identifier as the absorption boundary trigger criterion; The typical scenario generation module is used to perform feature clustering based on the historical operation data of the transformer area, couple discrete samples of line impedance, and generate a typical scenario set of source-load and parameter coupling. The initial boundary solution module is used to solve the initial absorption boundary value of each time section by employing the variable step size increment strategy and the AC / DC alternating iterative power flow algorithm. The confidence boundary correction module is used to fit the probability distribution function of the line impedance parameters, and to obtain the absorption boundary interval of multiple confidence levels and locate the absorption bottleneck constraint through probability sampling calculation. The protection verification output module is used to build a simulation model of the transformer area protection device and restore the protection action characteristics, perform steady-state over-limit and timing linkage verification, correct the boundary values, and output the final absorption boundary curve and bottleneck location results.

Images