A distributed photovoltaic bearing capacity analysis and evaluation method

Abstract

The application discloses a distributed photovoltaic bearing capacity analysis and evaluation method, relates to the technical field of power distribution network planning, and comprises the following steps: generating a basic evaluation value based on the maximum bearing capacity of distributed photovoltaic under different operating states; generating a health index according to the temperature load curve and operation history of each key equipment; calculating a cross-level stress coefficient according to an iterative grading method; performing space-time alignment on the health index and the cross-level stress coefficient to obtain aging acceleration and cumulative aging amount; calculating a bearing capacity adjustment factor according to the aging acceleration and the cumulative aging amount; correcting the basic evaluation value according to a correction function according to a correction reference value to obtain a final evaluation value; through dynamic correlation of the health index and the cross-level stress coefficient, the hidden aging risk of equipment is accurately evaluated, and the bearing capacity prediction accuracy is improved; dynamic perception and early warning of cross-level interaction influence are realized, and optimistic deviation in equipment life evaluation is avoided.

Description

Technical Field

[0001] This invention relates to the field of power distribution network planning technology, and in particular to a method for analyzing and evaluating the carrying capacity of distributed photovoltaic power. Background Technology

[0002] In the context of high penetration of distributed photovoltaic systems, existing capacity assessment methods are mainly based on static simulation of macroscopic operating data (such as voltage, current, and load rate), ignoring the dynamic effects of microscopic aging of equipment and cross-level interaction. This often leads to optimistic biases in the assessment results and makes it impossible to warn of hidden risks.

[0003] In assessing the carrying capacity of distributed photovoltaic (PV) systems, there is a cross-level interaction between the distribution network and the transmission network. For example, PV reverse power flow can cause boundary power flow reversal, leading to overload or voltage instability in transmission network equipment (such as main transformers), and equipment aging further exacerbates this risk. Traditional assessment methods ignore the cross-level cumulative effect, meaning that PV integration not only affects the lifespan of local equipment but also accelerates the aging of transmission network equipment through interaction, thereby reducing the overall carrying capacity.

[0004] Existing analytical methods do not adequately identify the interaction between distribution networks and transmission networks, thus failing to accurately quantify the degree of interaction, leading to a decrease in the accuracy of photovoltaic carrying capacity analysis and assessment. Summary of the Invention

[0005] This application provides a distributed photovoltaic carrying capacity analysis and evaluation method, which solves the problem of optimistic bias in the evaluation caused by ignoring the micro-aging of equipment and the dynamic effects of cross-level interactions in the prior art. It achieves the technical effect of fully identifying and analyzing interactive effects, and improving the evaluation accuracy by combining micro and macro levels.

[0006] This application provides a method for analyzing and evaluating the carrying capacity of distributed photovoltaic power, including: S1: Generate basic assessment values ​​based on the maximum carrying capacity of distributed photovoltaic power under different operating conditions; pre-set time periods, acquire real-time operating data, determine key equipment groups, and generate health indices based on the temperature load curves and operating history of each key equipment; calculate cross-level stress coefficients according to the iterative grading method, where the cross-level stress coefficients are the ratio of boundary power flow reversal power to the rated capacity of the transformer. S2: Spatiotemporally align the health index and cross-level stress coefficient to obtain health index sequence data and coefficient sequence data; establish a quantitative relationship model, set the time step, and obtain the health index decay rate corresponding to the cross-level stress coefficient at each time step according to the quantitative relationship model; record the cumulative decay amount to obtain the aging acceleration and cumulative aging amount; calculate the bearing capacity adjustment factor according to the aging acceleration and cumulative aging amount. S3: Obtain the bearing capacity adjustment factor sequence, take the 5th percentile of the bearing capacity adjustment factor sequence as the correction benchmark value, and correct the basic evaluation value according to the correction function based on the correction benchmark value to obtain the final evaluation value; The real-time operational data includes distribution network operational data and transmission network operational data.

[0007] Furthermore, the method also includes: S4: Obtain the final evaluation value, health index sequence, cross-level stress coefficient, and safety constraint verification record for each photovoltaic node, and generate a comprehensive risk score; photovoltaic nodes with a comprehensive risk score greater than 0.6 are marked as high-risk nodes; S5: Obtain the micro data sequence of each high-risk node to get the micro health value; calculate the final health value by coupling the health index and micro health value of the same time period through the correction formula; correct the cross-level stress coefficient according to the instantaneous change rate of crack propagation rate to get the micro stress coefficient; obtain the micro coefficient according to the final health value and micro stress coefficient, and update the bearing capacity by adjusting the bearing capacity adjustment factor according to the micro coefficient.

[0008] Furthermore, the iterative hierarchical method is as follows: calculate the power flow of the transmission network, using the power of the boundary nodes as the input to the distribution network; calculate the power flow of the distribution network, feeding back the voltage of the boundary nodes to the transmission network; repeat until the boundary conditions converge. This involves setting the number of iterations and the starting point of each iteration, pre-setting a convergence threshold and several time windows, acquiring real-time running data and boundary node identifiers within each time window, and assigning initial power and initial voltage values ​​to the boundary nodes. Calculate the voltage distribution of transmission network nodes and the voltage of boundary nodes; set the voltage of boundary nodes as the reference value of distribution network boundary nodes and use it as a boundary condition; obtain the voltage distribution of distribution network nodes and the power of boundary nodes based on real-time operating data; use the boundary node power calculated from the distribution network as the power injection value of the transmission network boundary nodes for the next calculation to update the transmission network power flow; after the calculation is completed, determine the deviation value based on the boundary node power of the two calculations; compare the deviation value with the convergence threshold; if it is less than the convergence threshold, output the steady-state power flow data and the final boundary node power, and mark the final boundary node power as the boundary reversal power; if it is greater than the convergence threshold, continue the iterative calculation.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: By quantifying the dynamic correlation between health indices and cross-level stress coefficients, the hidden aging risks of equipment are accurately assessed, improving the accuracy of load-bearing capacity prediction. Based on micro-parameter correction of macro-models, dynamic perception and early warning of cross-level interactive effects are achieved. By integrating multi-source sensor data to construct an aging acceleration index, optimistic biases in equipment life assessment are avoided, making the load-bearing capacity boundary closer to the actual safety threshold. By adaptively adjusting factors to optimize the load-bearing capacity assessment results in real time, the lag of static models is avoided, maximizing the safe utilization rate of photovoltaic access capacity. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of a distributed photovoltaic carrying capacity analysis and evaluation method in an embodiment of the present invention. Detailed Implementation

[0011] To facilitate understanding of the present invention, a more complete description of this application will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the present invention.

[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0013] This embodiment quantifies the data relationship between cross-level interaction, equipment lifespan, and load-bearing capacity. By integrating real-time monitoring, aging prediction, and dynamic assessment, it achieves more accurate and safer load-bearing capacity analysis. While improving the accuracy of load-bearing capacity analysis and assessment, it couples equipment health status with cross-level power flow, dynamically adjusting the upper limit of load-bearing capacity with lifespan decay as a constraint.

[0014] Example 1: As Figure 1 As shown, a method for analyzing and evaluating the carrying capacity of distributed photovoltaic power generation includes: S1: Generate basic assessment values ​​based on the maximum carrying capacity of distributed photovoltaic (PV) under different operating conditions. Specifically, pre-set safe operation constraints, obtain the maximum carrying capacity of distributed PV under different operating conditions, and calculate the upper limit of PV access capacity for each operating condition (e.g., season, load level) through iterative simulation. For the current operating condition (e.g., summer peak load), initialize the distributed PV access capacity of all PV access nodes to zero (i.e., assume no PV access). Simultaneously, set the capacity growth step size (e.g., 0.1MW) and the maximum number of iterations (e.g., 100 times) to prevent infinite loops. Gradually increase the PV capacity to simulate the process of increasing PV penetration. Based on the distributed PV development coefficient (calculated through GIS satellite imagery), increase the access capacity of each node proportionally. The growth rate is the product of the capacity growth step size and β (e.g., if node β=0.5, step size 0.1MW, then the current increase is 0.05MW), making the capacity growth more consistent with actual rooftop distribution, avoiding idealized uniform growth, and updating the PV access capacity of each node.

[0015] The system verifies whether the increased capacity leads to grid parameters exceeding limits based on safe operation constraints, ensuring the safety and reliability of the assessment results. These constraints include: branch current constraints (branch current does not exceed the long-term allowable current carrying capacity); node voltage constraints (node ​​voltage is within the allowable deviation range); and power balance constraints (active and reactive power balance equations are valid). If all constraints are satisfied, the system returns to step S12 to continue increasing the capacity; otherwise, the iteration stops, and the sum of the capacities at the last satisfied constraint is taken as the maximum carrying capacity under this operating state. Finally, the maximum carrying capacity of distributed photovoltaic power under a single operating state is obtained.

[0016] From all operating states, the most conservative carrying capacity value is selected to ensure safety throughout the year. The minimum maximum carrying capacity of distributed photovoltaic power generation and its corresponding operating state are selected to determine whether the safe operation constraints are met. If they are met, the maximum carrying capacity of distributed photovoltaic power generation is used as the basic evaluation value; if not, a re-evaluation is conducted.

[0017] Pre-set time periods, acquire real-time operating data, identify key equipment groups, and generate health indices based on the temperature load curves and operating history of each key equipment; The real-time operational data includes distribution network operational data and transmission network operational data. The distribution network operational data includes photovoltaic output, node voltage, and branch current, while the transmission network operational data includes main transformer load rate, voltage data, and ambient temperature. Simultaneously, IoT sensors (such as temperature sensors and vibration sensors) are deployed to monitor the health parameters of key equipment (such as transformers and lines), including winding temperature, insulation resistance, and operating time.

[0018] The acquired distribution network and transmission network operation data are converted into a standard format using a unified communication protocol, and timestamped to eliminate format differences between different systems. This generates a unified data view, ensuring that boundary node parameters (such as power exchange) are cross-verifiable. The merged data, i.e., real-time operation data, includes time-synchronized transmission and distribution network operation parameter tables, such as tables updated every 5 minutes, containing fields such as timestamp, transmission network transformer load rate, distribution network boundary node power, and equipment temperature.

[0019] The determination of key equipment groups needs to be based on the specific facilities and lines of the power grid. Selection is made based on the distribution of the lines and expert knowledge and experience, which will not be elaborated upon here. Historical data for each key piece of equipment is acquired and fitted to obtain a temperature-load curve. For example, transformer operating data is collected, including load rate (percentage), winding temperature (degrees Celsius), and ambient temperature (degrees Celsius), with a sampling interval of 1 hour. Ambient temperature is normalized to eliminate environmental influences. The formula is: Corrected temperature = Measured winding temperature - Ambient temperature × Compensation coefficient (the compensation coefficient is set to [0.5, 0.8], based on equipment characteristics; an example is 0.6). A linear regression is performed between the corrected temperature and the load rate to obtain the temperature-load curve equation: Corrected temperature = a × Load rate + b. Here, a and b are regression coefficients, obtained by fitting historical data using the least squares method. Assuming that after fitting the historical data, a = 0.5 and b = 20, the curve equation is: Corrected temperature = 0.5 × Load rate + 20. This means that for every 1% increase in load rate, the corrected temperature increases by 0.5°C.

[0020] Based on the real-time operating data of each device, the cumulative operating time and equivalent aging amount of each device are obtained. The equivalent aging amount is calculated by converting the operating history into standard aging units based on the temperature load curve. The formula is: Equivalent aging amount = Cumulative operating time × Aging rate coefficient. The aging rate coefficient is temperature-related and is commonly represented by a simplified version of the Arrhenius equation: Aging rate coefficient = exp(-Ea / (R×T)), where Ea is the activation energy (0.8 eV in the example), R is the gas constant (8.314 J / mol·K), and T is the absolute temperature (unit: K, taken from the temperature load curve). For simplification, a linear approximation can be made: Aging rate coefficient = k × Corrected temperature, where k is the device constant (in the example, k = 0.001 / °C·hour). First, calculate the average corrected temperature based on the temperature load curve. For example, if the transformer's annual average load rate is 60%, substituting into the curve equation: Corrected temperature = 0.5 × 60 + 20 = 50°C. Then calculate the aging rate coefficient: Aging rate coefficient = 0.001 × 50 = 0.05 / hour. Equivalent aging amount D = aging rate coefficient × cumulative operating time t.

[0021] The health index is calculated with an initial health index of 1, which approaches 0 with aging. The formula is: HI = 1 - 0.01 × D, where HI is the health index; D is the equivalent aging amount; and the cumulative aging rate is a characteristic parameter of the equipment, representing the rate of decline in the health index caused by a unit of equivalent aging amount, which is taken as 0.01 (i.e., the health index decreases by 0.01 per unit of equivalent aging amount).

[0022] For example, by determining the cumulative operating time t, obtaining the average corrected temperature from the temperature load curve, calculating the aging rate coefficient and equivalent aging amount, and then calculating the health index, a safety threshold is pre-set. The calculated health index is compared with the safety threshold; if the health index is less than the safety threshold, an early warning is issued. The health index quantifies equipment aging through the temperature load curve and operating history, providing dynamic constraints for load-bearing capacity assessment. This enhances the reliability of cross-level analysis and avoids overestimation of load-bearing capacity due to equipment aging.

[0023] The cross-level stress coefficient is calculated using an iterative grading method. This cross-level stress coefficient is the ratio of the boundary power flow reversal power to the transformer's rated capacity (e.g., the stress coefficient increases by 0.1 for every 10% increase in reversal power). It is used to quantify the intensity of the interaction between the distribution network and the transmission network caused by photovoltaic integration, reflecting the additional stress on key equipment (such as transformers) due to power flow reversal, i.e., the load pressure on transformers caused by power flow reversal. Its core purpose is to transform the abstract power flow impact into a quantifiable equipment stress index, thereby incorporating cross-level risk into the load-bearing capacity assessment.

[0024] The cross-level stress coefficient is a dimensionless ratio, and the formula is: ,in, It is the stress coefficient across different levels; This refers to the power transmitted from the distribution network to the transmission network in the reverse direction. When the power flow direction is opposite to the normal direction (power transmission from the distribution network to the distribution network), this power is negative. However, in calculations, the absolute value is taken or a negative value is used directly to represent the direction, depending on the application scenario. For example, if the boundary node power is -5MW, it means that the distribution network is transmitting 5MW of power back to the transmission network. This refers to the rated capacity of the transformer, specifically the rated capacity of the main transformer (such as the high-voltage / low-voltage side transformer in a substation) at the connection point between the transmission and distribution networks. Transformers are key equipment for cross-level interaction, and their capacity limits the permissible range of power flow reversal. For example, if σ=0.1, it means that the reverse power accounts for 10% of the transformer capacity. The larger the coefficient value, the higher the risk of equipment overload, which needs to be conservatively handled in the load-bearing capacity assessment.

[0025] The iterative hierarchical method is as follows: calculate the power flow of the transmission network, using the power of the boundary nodes as the input of the distribution network; calculate the power flow of the distribution network, feeding back the voltage of the boundary nodes to the transmission network; repeat until the boundary conditions converge.

[0026] Specifically: Set the number of iterations and the starting point of the iterations, pre-set the convergence threshold and several time windows, and obtain the real-time operating data (initial data of the transmission network and initial data of the distribution network) and boundary node identifiers within each time window; the boundary nodes are the transmission and distribution connection points, such as the low-voltage side of the substation, which are identified as interaction interfaces; and assign initial power values ​​and initial voltage values ​​to the boundary nodes. The process involves obtaining the current boundary conditions and time window, and solving for the power distribution and node voltages within the transmission network based on real-time operational data. Specifically, it uses the transmission network topology parameters, namely the node admittance matrix and branch impedance; obtains generator output data and load power based on operational data; and combines the power values ​​of boundary nodes to solve the power flow equations of the transmission network using power flow calculation methods. For each node in the transmission network, it solves the active and reactive power balance equations. For example, the power balance equation for node i is: injected power minus load power equals the sum of power flowing out from adjacent nodes (involving a combination of trigonometric functions of voltage, conductance, and susceptance). After the calculation is completed, the voltage magnitude and phase angle of all nodes in the transmission network are obtained, and the voltage distribution of the transmission network nodes and the voltage of the boundary nodes are output.

[0027] The voltages of boundary nodes are set as reference values ​​and used as boundary conditions. Power distribution and node voltages within the distribution network are solved based on real-time operational data. Specifically, based on distribution network topology parameters (node ​​admittance matrix and branch parameters), photovoltaic output data and load power are obtained from operational data. The distribution network power flow equations are solved using distribution network power flow calculation methods, combined with boundary node voltages. For each distribution network node, the power balance equation is solved, considering both photovoltaic power injection and load absorption power. For example, the equation for node j includes photovoltaic active power injection minus load active power, and then minus branch power losses. After calculation, the voltages and power of all nodes in the distribution network are obtained, and the distribution network node voltage distribution and boundary node power are output.

[0028] The boundary node power calculated from the distribution network is used as the power injection value for the transmission network boundary nodes in the next calculation to update the transmission network power flow. After the calculation, the deviation value is determined based on the boundary node power between the two iterations. The deviation value is compared with the convergence threshold. If it is less than the convergence threshold, the steady-state power flow data and the final boundary node power are output, and the final boundary node power is marked as the boundary reversal power. The steady-state power flow data is the power flow result of the transmission and distribution networks in the last iteration. If it is greater than the convergence threshold, the iterative calculation continues. Preferably, the voltage deviation needs to be checked simultaneously to ensure full convergence. When updating the transmission network power flow, the iteration loop is closed, allowing the transmission network to recalculate based on the new data to reflect the impact of the distribution network.

[0029] After the calculations are completed, data such as node voltages, branch power, and system losses from both the transmission and distribution networks are integrated. The final direction (positive or negative) of the boundary power flow and the load rate of key equipment are recorded to form a complete power flow report. Convergence depends on the deviation threshold of the boundary node power or voltage. Typically, the threshold is set to a small value (e.g., ε = 0.01 MW) to ensure accuracy. If convergence is not achieved after too many iterations (e.g., more than 100 iterations), it may indicate system instability, requiring data review or adjustment of initial conditions. The convergence of the iterative hierarchical method is generally good in practice because the coupling between transmission and distribution networks is relatively weak.

[0030] The iterative hierarchical method achieves efficient solution of transmission and distribution coordinated power flow by alternately calculating the power flow of the transmission and distribution networks and iteratively exchanging boundary data; it effectively handles cross-level problems caused by distributed photovoltaic access, provides reliable data for carrying capacity assessment, and ensures practicality and scalability.

[0031] S2: Spatiotemporally align the health index and cross-level stress coefficients to obtain health index sequence data and coefficient sequence data; establish a quantitative relationship model, set the time step, and obtain the health index decay rate corresponding to the cross-level stress coefficient at each time step according to the quantitative relationship model; specifically, based on the aligned sequence data, use a nonlinear regression model (such as an exponential decay model) to fit the relationship between the cross-level stress coefficient and the health index decay rate. The model input is the stress coefficient sequence, and the output is the health index decay rate. The time step is t, and the relationship is as follows: ,in, It is the rate of decay of the health index over time t; It is the cross-level stress coefficient (dimensionless) at time t. It is the health index at time t; These are all model parameters, obtained through fitting and training with historical data. Indicates the sensitivity to stress coefficient. Indicates the basic aging rate; (This is a nonlinear exponent, reflecting the stress acceleration effect). In this embodiment, based on historical data fitting and engineering practice, : Indicates the sensitivity of the stress coefficient, that is, for every unit increase in stress, the attenuation rate increases by 0.05. This represents the baseline aging rate; even without stress, the health index will decline at a rate of 0.01 per unit time. The nonlinear exponent reflects the stress acceleration effect, causing the stress influence to increase quadratically. Parameters are obtained through training with historical data: at least three years of equipment operation data are collected, and the least squares method or gradient descent is used to optimize the fitting formula, minimizing the mean square error between the predicted decay rate and the actual observed decay rate (calculated inversely through equipment condition assessment), iteratively adjusting until the error converges.

[0032] In some embodiments, the health index is a time-series value, acquired at different times within a set time period, such as daily or hourly. The initial value of the health index is set to 1 for each time period, which can be set to one year. The cross-level stress coefficient is the output of power flow simulation, corresponding to a specific operating scenario (such as peak photovoltaic output), and needs to be converted into a time series (e.g., by interpolating discrete simulation points into a continuous sequence). Using the timestamp as the key, the health index and cross-level stress coefficient are matched to the same time axis, resulting in health index sequence data and coefficient sequence data respectively. For example, if the health index is updated daily, while the stress coefficient is simulated hourly, the stress coefficient is aggregated to a daily granularity using a weighted average.

[0033] The cumulative decay amount is recorded to obtain the aging acceleration and the cumulative aging amount. The aging acceleration is the rate of change of the health index decay rate, reflecting the trend of the decay rate, and is calculated as follows: ,in, It is the acceleration of aging. It is the decay rate of the health index at time step t-1; This refers to the length of the time step, i.e., the length of the time window; the cumulative aging amount is the total decline in the health index within the corresponding time period, calculated using the following formula: ,in, It is a time period. It represents the cumulative aging amount, which increases with time and decay rate, initially starting at 0.

[0034] In cross-level interaction analysis, equipment aging and stress accumulation effects are key factors affecting the assessment of distributed photovoltaic carrying capacity. The health index quantifies the health status of the equipment (initially 1, decreasing with aging), while the cross-level stress coefficient reflects the intensity of interactive effects such as power flow reversal. Through spatiotemporal alignment and quantitative modeling, the health index decay rate, aging acceleration (the rate of change of the HI decay rate), and cumulative aging amount (the total decay amount of HI) are obtained.

[0035] The load-bearing capacity adjustment factor is calculated based on the aging acceleration and cumulative aging amount, using the following formula: ,in, It is the load-bearing capacity adjustment factor; The risk sensitivity coefficient is calibrated using equipment failure data, such as... The cumulative aging amount and aging acceleration cause the load-bearing capacity adjustment factor to decrease nonlinearly with aging and acceleration, with a value range of (0, 1). The absolute value of the aging acceleration ensures that the risk increases regardless of whether the acceleration is positive or negative, as both acceleration and deceleration may lead to instability.

[0036] The basic assessment value and bearing capacity adjustment factor in this application are both stable values ​​within a time period and need to be updated at different time periods; S3: Obtain the bearing capacity adjustment factor sequence, take the 5th percentile of the bearing capacity adjustment factor sequence as the correction benchmark value, and correct the basic assessment value according to the correction function based on the correction benchmark value to obtain the final assessment value.

[0037] In some embodiments, quantiles are statistical indicators used to describe the location of data distribution, dividing a data sequence into equal proportions. For example, the 5th percentile indicates that in the sorted bearing capacity adjustment factor sequence, 5% of the values ​​are less than or equal to that quantile, and 95% of the values ​​are greater than or equal to that quantile. It reflects the low-risk tail end of the data distribution and is often used to capture extremely conservative scenarios. The load capacity adjustment factor (LCAF) sequence is a dynamic value sequence calculated over a time step (e.g., per hour) (LCAF value range is (0,1]). Using the 5th percentile as the correction benchmark means selecting the maximum value among the lowest 5% of values ​​in the load capacity adjustment factor sequence (i.e., the 5th percentile value after sorting) as the correction reference benchmark. For example, if the LCAF sequence has 100 values, the 5th value after sorting (the 5th percentile position) is the correction benchmark value. This means that within the assessment period, the LCAF value is lower than this benchmark only 5% of the time, while the LCAF value is higher (lower risk) for the remaining 95% of the time. The 5th percentile corresponds to the worst-case scenario (such as accelerated equipment aging or periods of intense cross-level interaction). Correcting with this benchmark ensures that the assessment results are safe under all operating conditions, avoiding optimistic bias. It also smooths extreme fluctuations, reduces false alarms, and improves the robustness of the correction results.

[0038] The correction function is: ,in, This is the final evaluation value; This is the basic assessment value; It is a correction of the baseline value; The correction strength coefficient is calibrated using historical data. It is a calibration parameter (range [0,1]) used to control the correction magnitude of the bearing capacity adjustment factor deviation in the correction function. The larger the coefficient, the more aggressive the correction (the more conservative the final evaluation value R); the smaller the coefficient, the more minor the correction. For example, a value of 0.5 indicates a moderate correction strength, balancing risk and capacity. It can be preset through historical data analysis or obtained by fitting an optimization model, which will not be elaborated further in this application. These are adaptive weights with a value range greater than or equal to 1, used to amplify or reduce the correction effect based on the influence of aging acceleration. The calculation formula is as follows: , where 0.1 is the initial value of the sensitivity parameter, which can be dynamically adjusted according to the actual evaluation results within the time period; W is the absolute value of the average aging acceleration within the time period T; when the aging acceleration is large, W increases, strengthening the correction. A two-layer correction is formed based on the correction intensity coefficient and adaptive weights to avoid the dominance of a single factor and improve robustness.

[0039] During the assessment period, the carrying capacity adjustment factor and adaptive weights are recalculated every 24 hours, and the correction results are updated. If the rate of change of the correction results is lower than the threshold (e.g., 1%) for multiple consecutive times (e.g., 3 times), a convergence check is triggered, and the correction process is automatically terminated. Dynamic convergence is achieved through a feedback loop, reducing unnecessary calculations and adapting to changes in the power grid.

[0040] In distributed photovoltaic (PV) carrying capacity assessment, traditional methods calculate the maximum carrying capacity, i.e., the basic assessment value, based on static operating conditions. However, these methods neglect the spatiotemporal dynamics of equipment aging and cross-level interactions, potentially leading to overestimation or underestimation of the actual acceptable capacity. In this embodiment, a carrying capacity adjustment factor is calculated to correct the basic assessment value, effectively solving this problem and further improving the accuracy of PV carrying capacity assessment.

[0041] Example 2: In Example 1, a macroscopic assessment of the carrying capacity of distributed photovoltaics was achieved through multi-operation state iterative simulation and safety constraint verification. However, the above example relies on macroscopic data and cannot capture the early impact of microscopic defects in the equipment. This may cause the assessment results to lag behind the actual aging risks. In particular, under high-penetration photovoltaic access, the hidden risks at the microscopic level may trigger sudden failures, causing the carrying capacity assessment to deviate from the true safety boundary.

[0042] Due to the macroscopic limitations of traditional power equipment health assessments, the health index in distributed photovoltaic capacity assessments is usually calculated based on overall equipment parameters (such as temperature and load rate). However, equipment aging is essentially a gradual process from microscopic to macroscopic. For example, the aging of transformer insulation materials begins with molecular chain breakage, and winding deterioration begins with fatigue cracks in the metal lattice. These microscopic changes do not immediately manifest as abnormal macroscopic parameters in the early stages, but they accumulate over time and trigger sudden failures.

[0043] Changes at the microscopic level (such as molecular structure and lattice arrangement) cannot be captured by conventional monitoring in the early stages of equipment operation, but their cumulative effect can significantly reduce material properties (such as insulation strength and mechanical toughness). Traditional health indices rely on macroscopic data such as temperature and vibration. When these indicators become abnormal, the equipment has often entered an accelerated aging phase, causing the load-bearing capacity assessment results to deviate from the actual safety boundary. Existing models have not established a quantitative correlation between microscopic defects and macroscopic performance. For example, how molecular-level cracks affect winding thermal conductivity, thereby causing local overheating and ultimately manifesting as a decline in the overall health index.

[0044] This embodiment, through in-depth analysis of historical equipment failure cases (such as microscopic analysis of materials after transformer burnout), reveals that most failures originate from microscopic defects. For example, scanning electron microscopy of insulating paper samples shows that molecular chain breakage occurs earlier than macroscopic temperature anomalies, and the breakage density is positively correlated with load fluctuations. Breaking away from the framework of traditional electrical fault analysis, this embodiment introduces material failure analysis techniques, reconstructing the problem from "why the equipment failed" to "how microscopic defects evolved into failures." High-precision sensors are deployed on power equipment to achieve real-time acquisition of microscopic parameters. For example, ultrasonic sensors monitor changes in the sound wave propagation speed inside transformer windings, inferring the propagation rate of metal fatigue cracks.

[0045] S4: Obtain the final evaluation value, health index sequence, cross-level stress coefficient and safety constraint verification record for each photovoltaic node, and generate a comprehensive risk score; photovoltaic nodes with a comprehensive risk score greater than 0.6 are marked as high-risk nodes.

[0046] In some embodiments, all nodes are subjected to multi-index fusion screening, including: bearing capacity threshold screening, identifying nodes with insufficient bearing capacity based on the final evaluation value, obtaining the system average bearing capacity, and initially marking nodes whose bearing capacity value is lower than the system average bearing capacity, indicating that the node is a weak link in the global context and needs to be analyzed first.

[0047] Health index decay rate screening identifies points of accelerated aging, and microscopic analysis can provide early warnings of lifespan risks. If the health index decay rate exceeds 0.05 per month, meaning the health index decreases by more than 5% per month, it indicates that the equipment has entered an accelerated aging stage, requiring analysis of the microscopic root causes (such as crack propagation).

[0048] Cross-level stress coefficient screening selects nodes with high cross-level interaction risk, and microscopic analysis can verify the impact of stress on equipment. Nodes with cross-level stress coefficients greater than 0.15 (i.e., boundary power flow reversal power accounts for more than 15% of the transformer capacity) are selected.

[0049] Safety constraint screening identifies nodes that are close to exceeding limits during load capacity iteration. In the iterative verification of Example 1, if the voltage deviation or current load rate is close to the constraint limit (e.g., the voltage deviation reaches more than 90% of the allowable value), it indicates that the node is operating in a critical state.

[0050] Based on the above multi-indicator system, nodes are stratified and labeled. For labeled nodes (which will be labeled as long as they meet any one of the conditions), a comprehensive risk score is calculated by weighting the multi-indicator system, and a screening threshold (such as 0.6) is set. In the process of weighted summation, the weight of each indicator needs to be pre-set based on historical experimental data, which will not be elaborated on in this application.

[0051] For the list of high-risk nodes, S5: Obtain the micro data sequence of each high-risk node, decompose the original signal based on wavelet transform, extract the characteristic frequency components, calculate the molecular aggregation degree index and crack propagation rate, and then obtain the micro health value. A micro-macro coupling model is established based on long short-term memory networks to obtain the dynamic relationship between micro data sequences and real-time running data, and a coupling correction formula is generated. Based on the health index and micro-health value for the same time period, the final health value is calculated using a coupled correction formula, as follows: ,in, This is the final health value; It is a microscopic health value; It is a health index; It is the coupling coefficient, with a value range of [0.1, 0.2], which controls the correction strength.

[0052] The micro-stress coefficient is obtained by correcting the cross-level stress coefficient based on the instantaneous rate of change of crack propagation rate; the micro-coefficient is obtained based on the final health value and the micro-stress coefficient; the bearing capacity is then updated based on the micro-coefficient to obtain the updated bearing capacity adjustment factor. The calculation formula is as follows: ,in, It is the micro-stress coefficient; It is the cross-level stress coefficient, calculated based on the above embodiments; It is the instantaneous rate of change of the crack propagation rate, and the formula is: ,in, t is the crack propagation rate, and t is the time window; This is the expansion factor, used to control the intensity of microscopic effects; it is set to 0.2 here. It is the temperature of a local hot spot; This refers to the rated temperature of the equipment. When analyzing a single device, the temperature is directly obtained using a sensor. However, when analyzing multiple critical devices together, the average hotspot temperature of the critical devices can be taken as the local hotspot temperature. The selected devices should be of the same type to ensure that the rated temperatures of the devices are consistent.

[0053] The principle behind the microscopic effects lies in the fact that aging acceleration reduces the equipment's stress tolerance, making it more prone to overload under the same power flow. Regression analysis of historical failure data revealed a positive correlation between the instantaneous rate of change of crack propagation rate and the failure probability. The propagation coefficient was obtained through least squares fitting (e.g., ...). When the prediction error is minimized.

[0054] In some embodiments, the purpose of establishing dynamic relationships is to create a nonlinear mapping model between microscopic parameters (such as crack propagation rate and molecular vibrational frequency) and macroscopic health indices. Traditional health index calculations rely solely on macroscopic data (such as temperature and load rate), but microscopic parameters can reflect equipment aging trends earlier (e.g., molecular-level degradation precedes temperature increases). By training machine learning models (such as LSTM), the complex relationship between the microscopic and macroscopic levels can be captured, enabling accurate prediction of health indices and thus providing early warnings of hidden risks.

[0055] The formula for generating micro coefficients is as follows: ,in, These are micro-level coefficients; These are risk weighting coefficients, recalibrated using historical data (e.g., To reflect the sensitivity of micro-level risks, micro-level coefficients are directly integrated into the carrying capacity adjustment factor, enabling the carrying capacity assessment to simultaneously respond to changes in macro-level operating conditions and micro-level defects, thus improving accuracy. Specifically, micro-level coefficients are integrated into the macro-level assessment to generate a comprehensive carrying capacity adjustment factor. A product method is used, where the product of the micro-level coefficients and the carrying capacity adjustment factor is used as the updated carrying capacity adjustment factor, ensuring that the combined effect of micro- and macro-level risks is linearly amplified. For example, when both macro-level and micro-level risks are high, the combined carrying capacity adjustment factor decreases significantly, indicating high overall risk; the value range remains (0,1], facilitating direct use for carrying capacity correction.

[0056] The real-time operating data and microscopic data sequences obtained in this application need to be acquired within a set time period. The microscopic data sequences include raw ultrasonic sensor signals (sampling frequency 1MHz, accuracy ±0.1μm) for monitoring metal crack propagation; terahertz spectral data (wavelength 0.1-10THz) for analyzing the molecular vibration frequency of insulating materials; and acoustic emission signals (frequency range 50-400kHz) for capturing microscopic defect mutation events.

[0057] In this embodiment, the limitations of traditional macroscopic assessment are overcome by integrating microscopic parameters into the load-bearing capacity assessment in real time. By training the micro-macroscopic dynamic relationship through an LSTM network, cross-scale coupling of equipment aging from the molecular level to the system level is realized. A microscopic stress coefficient correction formula is set to combine microscopic aging acceleration with hot spot temperature, quantifying the implicit risks of power flow interaction. A microscopic coefficient adjustment load-bearing capacity adjustment factor is introduced to make the assessment results respond to changes in operating conditions and microscopic defects simultaneously.

[0058] This technology addresses the long-standing issues in existing technologies, such as insufficient early warning of microscopic defects leading to assessment lags and missed risk detection, as well as the inability of macroscopic models to capture the profound impact of cross-level interactions. It reduces health index prediction errors and carrying capacity assessment biases, advances the warning time for latent faults, and improves cross-level risk coverage, thereby significantly enhancing the safety and reliability of the power grid.

[0059] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for analyzing and evaluating the carrying capacity of distributed photovoltaic power, characterized in that, include: S1: Generate basic assessment values ​​based on the maximum carrying capacity of distributed photovoltaic power under different operating conditions; Pre-set time periods, acquire real-time operating data, identify key equipment groups, and generate health indices based on the temperature load curves and operating history of each key equipment; The cross-level stress coefficient is calculated according to the iterative classification method. The cross-level stress coefficient is the ratio of the boundary power flow reversal power to the rated capacity of the transformer. S2: Spatiotemporally align the health index and cross-level stress coefficient to obtain health index sequence data and coefficient sequence data; Establish a quantitative relationship model, set the time step, and obtain the health index decay rate corresponding to the stress coefficient across levels at each time step based on the quantitative relationship model; Record the cumulative attenuation to obtain the aging acceleration and cumulative aging amount; calculate the load-bearing capacity adjustment factor based on the aging acceleration and cumulative aging amount; S3: Obtain the bearing capacity adjustment factor sequence, take the 5th percentile of the bearing capacity adjustment factor sequence as the correction benchmark value, and correct the basic evaluation value according to the correction function based on the correction benchmark value to obtain the final evaluation value; The real-time operational data includes distribution network operational data and transmission network operational data.

2. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 1, characterized in that, The method further includes: S4: Obtain the final evaluation value, health index sequence, cross-level stress coefficient, and safety constraint verification record for each photovoltaic node, and generate a comprehensive risk score; photovoltaic nodes with a comprehensive risk score greater than 0.6 are marked as high-risk nodes; S5: Obtain the micro data sequence of each high-risk node to get the micro health value; calculate the final health value by coupling the health index and micro health value of the same time period through the correction formula; correct the cross-level stress coefficient according to the instantaneous change rate of crack propagation rate to get the micro stress coefficient; obtain the micro coefficient according to the final health value and micro stress coefficient, and update the bearing capacity by adjusting the bearing capacity adjustment factor according to the micro coefficient.

3. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 1, characterized in that, The iterative hierarchical method is as follows: calculate the power flow of the transmission network, using the power of the boundary nodes as the input to the distribution network; calculate the power flow of the distribution network, feeding back the voltage of the boundary nodes to the transmission network; repeat until the boundary conditions converge. This involves setting the number of iterations and the starting point of each iteration, pre-setting a convergence threshold and several time windows, acquiring real-time running data and boundary node identifiers within each time window, and assigning initial power and initial voltage values ​​to the boundary nodes. Calculate the voltage distribution of transmission network nodes and the voltage of boundary nodes; set the voltage of boundary nodes as the voltage reference value of distribution network boundary nodes and use it as the boundary condition; obtain the voltage distribution of distribution network nodes and the power of boundary nodes based on real-time operating data. The boundary node power obtained from the distribution network calculation is used as the power injection value of the transmission network boundary node for the next calculation to update the transmission network power flow; after the calculation is completed, the deviation value is determined based on the boundary node power of the two calculations. The deviation value is compared with the convergence threshold. If it is less than the convergence threshold, the steady-state power flow data and the final boundary node power are output, and the final boundary node power is marked as the boundary reversal power. If it is greater than the convergence threshold, the iterative calculation continues.

4. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 1, characterized in that, Calculating the health index includes: based on the real-time operating data of each device, obtaining the cumulative operating time and equivalent aging amount of each device. The health index is initially set to 1 and approaches 0 as aging occurs. The calculation formula is as follows: Where HI is the health index; D is the equivalent aging amount; and 0.01 is the cumulative aging rate, which represents the rate of decline in the health index caused by a unit of equivalent aging amount.

5. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 1, characterized in that, To obtain the health index decay rate corresponding to the cross-level stress coefficient at each time step, the following steps are performed: Based on aligned sequence data, a nonlinear regression model is used to fit the relationship between the cross-level stress coefficient and the health index decay rate. The model input is the stress coefficient sequence, and the output is the health index decay rate. The time step is t, and the relationship is as follows: ； in, It is the rate of decay of the health index over time t; It is the cross-level stress coefficient at time t; It is the health index at time t; These are all model parameters, obtained through fitting and training with historical data. Indicates the sensitivity to stress coefficient. Indicates the basic aging rate; It is a nonlinear exponent that reflects the stress acceleration effect.

6. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 1, characterized in that, The load-bearing capacity adjustment factor is calculated based on the aging acceleration and cumulative aging amount, including: the aging acceleration is the rate of change of the health index decay rate, reflecting the changing trend of the decay rate, as shown in the following formula: ； in, It is the acceleration of aging. It is the decay rate of the health index at time step t-1; It is the length of the time step; The cumulative aging amount is the total decline in health index within the corresponding time period, and the formula is: ； in, It is a time period. It represents the cumulative aging amount, which increases with time and decay rate, initially starting at 0. ； in, It is the load-bearing capacity adjustment factor; The risk sensitivity coefficient is calibrated using equipment failure data. .

7. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 1, characterized in that, The correction function is: ； in, This is the final evaluation value; This is the basic assessment value; It is a correction of the baseline value; It is a corrected intensity coefficient, calibrated using historical data, with a value range of [0, 1]. These are adaptive weights with a value range greater than or equal to 1, used to amplify or reduce the correction effect based on the influence of aging acceleration. The calculation formula is as follows: ； Where 0.1 is the initial value of the sensitivity parameter; It is the absolute value of the average aging acceleration within the time period T.

8. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 2, characterized in that, The final health value is obtained by: acquiring the microscopic data sequence of each high-risk node, decomposing the original signal based on wavelet transform, extracting characteristic frequency components, calculating the molecular aggregation degree index and crack propagation rate, and thus obtaining the microscopic health value; establishing a micro-macro coupling model based on a long short-term memory network, obtaining the dynamic relationship between the microscopic data sequence and real-time running data, and generating a coupling correction formula; and calculating the final health value using the coupling correction formula. The formula is as follows: ； in, This is the final health value; It is a microscopic health value; It is a health index; It is the coupling coefficient, with a value range of [0.1, 0.2], which controls the correction strength.

9. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 2, characterized in that, The micro-stress coefficient is obtained from the following formula: ,in, It is the micro-stress coefficient; It is the stress coefficient across different levels; It is the instantaneous rate of change of the crack propagation rate, and the formula is: ,in, t is the crack propagation rate; t is the time window. This is the expansion factor, used to control the intensity of microscopic effects, set to 0.2; It is the temperature of a local hot spot; This is the rated temperature of the equipment.

10. The distributed photovoltaic carrying capacity analysis and evaluation method as described in claim 2, characterized in that, The formula for generating micro coefficients is as follows: ； in, These are micro-level coefficients; It is a risk weighting coefficient, recalibrated using historical data. This reflects the sensitivity to micro-risks; the product of the micro-coefficient and the bearing capacity adjustment factor is used as the updated bearing capacity adjustment factor.

Images