Distribution transformer electric vehicle charging load dynamic capacity and risk assessment method

Abstract

The invention discloses a distribution transformer electric vehicle charging load dynamic capacity and risk assessment method. The method comprises the following steps: acquiring multi-dimensional diagnosis data of a transformer to construct a comprehensive health index, and mapping the comprehensive health index into a derating coefficient; the method comprises the following steps: predicting a basic load curve by using a Holt-Wittes prediction basic load curve, aggregating load characteristic parameters of batch charging piles based on a coincidence factor, constructing a total load of a transformer, correcting rated temperature rise parameters of a hot-spot temperature response model by using a health derating coefficient, and constructing a temperature probability density by using moment operation and Gramer-Networker expansion; and finally, iteratively adding the charging piles and calculating the over-temperature risk probability until a set threshold value is reached, and determining the maximum accessible number. According to the method, the equipment health state and the load randomness are considered, dynamic capacity accurate evaluation is realized, and a quantitative basis is provided for safe access of charging facilities.

Description

Technical Field

[0001] This invention relates to the field of power system distribution network operation and asset management technology, specifically to a method for assessing the dynamic capacity and risk of electric vehicle charging load on distribution transformers. Background Technology

[0002] With the acceleration of global energy transition and transportation electrification, electric vehicles, as a new type of high-power load, have brought unprecedented challenges to the power distribution network due to their large-scale integration. As a key node connecting medium-voltage and low-voltage networks, the remaining carrying capacity of distribution transformers directly determines the scale of charging loads that can be accommodated within a region. However, existing assessment techniques have significant limitations: methods based on static ratings ignore the performance degradation of transformers due to insulation aging, oil deterioration, and overload history, resulting in overly conservative results and potential resource waste; deterministic worst-case analysis assumes all loads operate at maximum power simultaneously, failing to consider the randomness of user charging behavior, leading to an underestimation of the available load and a lack of economic rationality; while black-box models based on machine learning and probabilistic methods based on Monte Carlo simulations can improve accuracy or reflect uncertainty, they suffer from poor interpretability, high data requirements, and large computational loads, making real-time application difficult, respectively.

[0003] In summary, existing methods exhibit an imbalance between simplified and complex models. Therefore, an evaluation method is urgently needed to address these issues. Summary of the Invention

[0004] The technical problem to be solved by this invention is how to address the issues of a one-size-fits-all approach to safety margins and the inability to quantify risks in assessment results in traditional methods.

[0005] This invention solves the above-mentioned technical problems through the following technical means: a method for assessing the dynamic capacity and risk of electric vehicle charging load in distribution transformers, comprising: S1. Collect multi-dimensional diagnostic data of the target transformer, construct a comprehensive health index, and transform the comprehensive health index into a health depreciation coefficient through non-linear mapping; S2. Based on the historical load data of the transformer, the basic load curve for future periods is obtained by using the time series decomposition method, and the load characteristic parameters of batch charging piles are aggregated based on the overlap factor. S3. The total load of the transformer is obtained by superimposing the deterministic base load and the random charging load. After correcting the rated temperature rise parameter of the hot spot temperature response model with the health derating factor, the first four central moments of the hot spot temperature are obtained by analytical calculation of the total load statistical moments through moment operation. Based on this, the probability density function of the hot spot temperature is constructed by the Gram-Charlie series expansion method. S4. Based on the hotspot temperature probability distribution obtained in step S3, set a probabilistic safety constraint for hotspot temperature exceeding the limit; adjust the number of charging piles by incremental iteration method, and recalculate the hotspot temperature probability distribution under the new number for safety verification, determine the maximum number of charging piles that can be connected, and output the margin that can be added.

[0006] This invention constructs a comprehensive health index by collecting multi-dimensional diagnostic data of transformers and maps it to a health derating coefficient, thereby correcting the rated parameters of the hotspot temperature model. It employs time series decomposition and overlapping factor aggregation to characterize the base load and the random charging load of batch charging piles, respectively. After superposition, it uses moment calculation and Gram-Charlieer series expansion to obtain the complete probability distribution of hotspot temperature. Based on this, it sets a temperature exceedance probability constraint and dynamically verifies the thermal risk under different charging pile access scales using an incremental iteration method. This accurately outputs the maximum number of charging piles that can be connected and the margin for new additions, enabling the full exploitation of the charging service capacity without compromising transformer lifespan safety. This significantly improves the distribution network's elasticity in accepting highly random charging loads and enhances asset utilization efficiency.

[0007] Furthermore, step S1 specifically includes: S1.1 Construction and Quantification of Comprehensive Health Index: Collect multi-dimensional diagnostic data of the target transformer, including dissolved gas analysis, insulating oil quality, insulating paper aging parameters, historical load and maintenance records. Construct the comprehensive health index HI value by standardizing the multi-dimensional diagnostic data and weighted summing. S1.2, Health Deduction Coefficient: Defines the functional health deduction coefficient of HI. A piecewise nonlinear function is used to reflect the nonlinear characteristics of the rate of deterioration of health status; The segmentation points of the piecewise nonlinear function are based on the HI health level classification: .

[0008] Furthermore, in step S2, the time series decomposition method is the Holt-Winters exponential smoothing method. This method decomposes historical daily load data into horizontal, trend, and seasonal components. The calculation formula is as follows:

[0009] in, It is the observed load value at time point t. , , These are the level, trend, and seasonal components at time point t. This represents the predicted load value at time point t for a future time step h. , , This is the smoothing parameter, ranging from 0 to 1; m is the length of the seasonal cycle; h is the number of forward prediction steps; k = [(h...]. 1) / m], after optimization of smoothing parameters, extrapolation yields the deterministic base load curve for the next 24 hours.

[0010] Furthermore, the formula for calculating the relationship between the overlap factor and the number of charging piles mentioned in step S2 is as follows:

[0011] in, , , These are the fitting parameters. This refers to the number of charging stations.

[0012] Furthermore, the load characteristic parameters of the batch charging piles aggregated based on the overlap factor are specifically as follows: Peak polymer power:

[0013] Average polymerization load:

[0014] Aggregate load variance:

[0015] in, This is the rated power of a single charging station. This represents the probability that a single EV is in a charging state at time t.

[0016] Furthermore, the method of superimposing the deterministic base load and the stochastic charging load to obtain the total transformer load specifically involves: The formula for calculating the total load of a transformer is:

[0017] in, For deterministic base loads, For random charging loads, the statistical moments of the total load are: the mean is ,variance skewness kurtosis .

[0018] Furthermore, the hotspot temperature response model is specifically as follows: Transformer hot spot temperature response model:

[0019] Winding hot spot temperature rise equation:

[0020] The final hotspot temperature is: ,in It is the ambient temperature. and These represent the temperature rise of the top oil relative to the environment and the temperature rise of the hot spot relative to the top oil, respectively. For load factor, and These are the top oil temperature rise and hot spot temperature rise under the corrected rated load, respectively. It is the ratio of rated load loss to no-load loss. and These are the thermal time constants of the oil and the winding, respectively. and It is an experience index.

[0021] Furthermore, the formula for calculating the rated temperature rise parameter of the hotspot temperature response model using the health derating factor is as follows:

[0022]

[0023] in, The top oil temperature rise under rated load, This refers to the temperature rise of the hot spot under rated load.

[0024] Furthermore, the probability density function for constructing the hotspot temperature using the Gramm-Charlieel series expansion method is specifically as follows: The hotspot temperature is standardized, and its calculation formula is as follows:

[0025] in, The standardized hotspot temperature, Hotspot temperature The statistical mean, Hotspot temperature Standard deviation; probability density function Approximation via the following series:

[0026] in, It is the probability density function of the standard normal distribution. and They are The skewness and excess kurtosis are calculated from its third and fourth moments. It is the first nHermitian polynomial of order 1.

[0027] Furthermore, S4 specifically includes: S4.1 Setting safety constraints: At any time during the evaluation period, the transformer winding hot spot temperature Exceeding the long-term operating temperature limit of the insulating material The probability is less than the risk threshold set by the grid operator. Its mathematical expression is:

[0028] Wherein, this probability By using the probability density function of standardized hotspot temperature The calculation is performed by integration, and the formula is as follows:

[0029] S4.2 Calculate the maximum number of new charging piles using the incremental iteration method. The specific steps are as follows: (1) Initialization: Let the number of currently connected charging piles be . The iteration increment i=1; (2) Iterative loop: a. Set the total number of charging piles to be evaluated. ; b. Perform preceding calculations: Using the number of charging piles as input, the calculation process in steps S2 and S3 is executed to obtain the result. Given a quantity, evaluate the probability density function of hotspot temperature at each moment within the evaluation period. ; c. Risk probability test: Calculate the risk probability at each moment within the assessment period. Find the moment when the risk probability is the highest, and determine whether the highest risk probability satisfies the safety constraint. If it satisfies the safety constraint, continue to step d; otherwise, proceed to step e. d. Incremental iteration: Record the current... Let i be the acceptable number of charging stations; let i = i + 1, and return to step a for the next iteration; e. Terminate iteration: If the safety constraints are not met, then The number of charging piles has exceeded the safety margin, and the iteration process has been terminated. (3) Determine the result: The maximum total number of securely accessible devices is ; (4) Output margin: The final margin that can be added is ; S4.3 Output the final evaluation results: The output includes the basic information of the transformer and the margin that can be added.

[0030] The advantages of this invention are: 1. By introducing a comprehensive health index and a dynamic depreciation mechanism, this invention breaks through the limitations of traditional static nameplate capacity, enabling the assessment results to truly reflect the individual differences and time-varying health status of transformers, thereby avoiding the huge errors caused by a one-size-fits-all assessment and significantly improving the level of precision in asset utilization.

[0031] 2. This invention replaces traditional deterministic thresholds with probabilistic safety constraints, enabling the reliability of assessment results to be quantified using explicit probabilistic language. This risk-based decision-making method allows grid operators to make scientific and quantitative trade-offs between safety and economic benefits.

[0032] 3. The entire evaluation process of this invention consists of a series of mathematical formulas and physical models. This not only makes the calculation extremely fast, but also ensures that every step from input data to the final conclusion is clear, transparent, traceable, and verifiable, meeting the stringent requirements of power system safety-critical applications. Attached Figure Description

[0033] Figure 1 This is a flowchart of the dynamic capacity and risk assessment method for electric vehicle charging load of distribution transformer according to Embodiment 1 of the present invention; Figure 2 This is a diagram showing the calculation results of Embodiment 1 of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, 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.

[0035] Example 1 like Figure 1 As shown, the dynamic capacity and risk assessment method for electric vehicle charging load of distribution transformers includes: S1. Transformer health status quantification and dynamic derating of baseline capacity Specifically, unlike the traditional view that all transformers are homogeneous and have constant performance, this approach deeply mines their operation and maintenance data to create a precise and quantitative health profile for each assessment object, and directly applies it to subsequent heat capacity assessments.

[0036] S1.1 Construction and Quantification of the Comprehensive Health Index: First, a quantitative and comprehensive Health Index (HI) is established to assess the current actual health status of the transformer. This index is obtained by standardizing and weighting multi-dimensional diagnostic data. This data mainly includes, but is not limited to: dissolved gas analysis, insulating oil quality, insulating paper aging degree (furan analysis), historical load, power factor, and other maintenance records. The general formula for calculating the comprehensive health index is as follows:

[0037] in, It is the first The health index factor for each assessment item (such as DGA, oil quality, etc.) is a standardized score that converts the raw measurements according to industry standards (IEEE C57.104). These are the corresponding weighting coefficients, reflecting the contribution of different evaluation items to the overall health of the transformer. This method transforms complex equipment status information into a single, comparable HI value.

[0038] S1.2 Health Derating Factor: The HI value will be directly converted into an operable parameter that can be used for engineering calculations. Transformers in poor health will have reduced internal heat dissipation capacity, and hot spots will accumulate more easily; therefore, their actual heat capacity should be lower than their nameplate value. A health derating factor is defined. To quantify this performance degradation. This is a function of HI (Health Degradation Factor), typically a non-linear function to reflect the physical reality that health conditions deteriorate more rapidly in the low to middle range. This invention represents the health degradation factor as a piecewise function, with the segmentation points referencing the HI health level classification: .

[0039] Reduction coefficient The key parameters of the hot spot temperature response model will be directly used in step S3 to incorporate the transformer's age and health status into the calculation of the thermal limit, thereby enabling accurate calculation of the distribution transformer's open capacity.

[0040] S2, Base Load and Charging Load Characteristic Modeling This step establishes a mathematical model that accurately describes the future total load characteristics of the transformer. This model must be able to separate the deterministic and stochastic components and provide an analytical description of the stochasticity.

[0041] S2.1, Time-series decomposition and prediction of basic load For existing base loads of transformers (mainly residential and commercial electricity), their daily load curves typically exhibit obvious periodicity (seasonality), trends, and horizontality. The Holt-Winters exponential smoothing method is used for load forecasting. This method effectively decomposes the time series and performs extrapolation forecasting. The additive model used is suitable for scenarios where the seasonal fluctuation amplitude does not change with the series level, which is consistent with the characteristics of residential daily load. The model consists of the following set of formulas:

[0042] in, At a certain point in time t The observed load value. , , They are time points t The level, trend, and seasonal components. This represents the predicted load value at time point t for a future time step h. , , It is a smoothing parameter, with a value between 0 and 1, which controls the weighting of recent observations. m This refers to the length of the seasonal cycle; for daily load curves, the sampling interval is 15 minutes. m =96. h It is the number of steps to predict forward. k =[( h 1) / m ] Parameters were adjusted using historical load data. , , Optimize to obtain the load forecast value with the minimum error. This will give you a base load curve that can predict the next 24 hours. P base ( t A deterministic model.

[0043] S2.2 Random Aggregation of Electric Vehicle Charging Load The charging behavior of a single EV is random, but the aggregated charging behavior of a large number of EVs exhibits statistical regularities on a macroscopic scale. This invention does not simulate individual EVs, but directly models the aggregated load mathematically. It employs the Coincidence Factor (CF), also known as the simultaneity rate factor. CF is defined as the ratio of the maximum total load of a group of users within a certain time period to the sum of the maximum loads of each user. Research shows that for EV charging load, CF increases with the number of charging stations. The increase leads to a significant decrease. This invention employs an analytical function validated through extensive data fitting to describe CF and Relationship:

[0044] in, , , These are the fitting parameters, and their values ​​are closely related to the charging scenario, such as the charging pile power and user charging habits. The CF function is used to calculate... Key statistical matrix of aggregated load of individual charging piles: (1) Peak power of polymerization:

[0045] in, This is the rated power of a single charging station. It can be calculated directly using the formula. The maximum load that a single charging station may generate is far less than the maximum load typically considered. .

[0046] (2) Average polymerization load:

[0047] in, Indicates a single EV at time [time]. t The probability of being in a charging state. This probability curve can be obtained from public datasets or local studies of typical charging behavior.

[0048] (3) Aggregate load variance: The variance of the aggregate load is fundamental for probabilistic analysis. The magnitude of the variance is related to both the peak and mean power, and is directly proportional to the difference between the mean and peak power. The following relationship can be established:

[0049] Assuming that aggregate loads follow a truncated normal distribution over a specific time period, their variance can be estimated by the mean and the upper limit of the confidence interval determined by the peak value.

[0050] The above method describes complex stochastic processes using their analytical statistical moments (mean, variance, peak value).

[0051] S3. Probabilistic thermal limit analysis considering uncertainties. The uncertainty of the load is transferred to the transformer hot spot temperature and described in a probabilistic form.

[0052] S3.1 Calculation of the characteristics of the total load stochastic process At any time t Total load on the transformer It is a deterministic base load and random charging load Superposition:

[0053] because Treated as a deterministic quantity (taking the predicted value), the statistical moments of the total load can be directly calculated according to the basic properties of probability theory: the mean is... ,variance skewness kurtosis The skewness and kurtosis of the total load are determined by the distribution characteristics of the charging load.

[0054] S3.2 Transformer Hot Spot Temperature Response Model

[0055] Winding hot spot temperature rise equation:

[0056] The final hotspot temperature is: in It is the ambient temperature. Among them, These represent the temperature rise of the top oil relative to the environment and the temperature rise of the hot spot relative to the top oil, respectively. K is the load factor, and R is the ratio of rated load loss to no-load loss. It is the thermal time constant of the oil and windings. x and y This is an empirical index, related to the cooling method. and These are the top oil temperature rise and hot spot temperature rise under the corrected rated load, respectively.

[0057] Based on the health reduction coefficient calculated in step S1 The rated temperature rise parameter is corrected:

[0058]

[0059] in, The top oil temperature rise under rated load, This step represents the temperature rise of hot spots under rated load. This correction is a crucial link between the transformer health status assessment and subsequent thermal risk calculations. Its core function is to transform the abstract health index HI assessed in step one into specific physical parameters that can be directly used by the thermodynamic model. All subsequent calculations will use this set of corrected temperature rise parameters that reflect the transformer's current true condition, rather than its idealized standard values ​​from the factory.

[0060] By correcting the temperature rise parameter, it will better reflect the physical changes in actual applications: a transformer in poor condition has reduced heat dissipation efficiency and will produce a higher temperature rise under the same load than a healthy transformer.

[0061] S3.3 Analytical Approximation of the Hotspot Temperature Probability Density Function The above thermal model equations are linear, and the load... The randomness can be linearly transferred to the output through these equations. Above. It can be analytically calculated. The statistical moments are obtained without stochastic simulation. This process is called the moment method, a classic analytical technique in the field of probabilistic load flow. By performing moment calculations on the thermal model equations, the statistical moments are obtained... The first four central moments: The probability density function is analytically constructed using the Gram-Charlier A-Series Expansion. This method uses a known distribution (usually a normal distribution) as a basis and approximates an unknown, nonnormal distribution by using higher-order moments (skewness and kurtosis) for correction.

[0062] First, the hotspot temperature is standardized: .in, The standardized hotspot temperature is a dimensionless standard fraction. Hotspot temperature The statistical mean (expected value). Hotspot temperature Standard deviation is used to quantify the magnitude of its fluctuation. Then, the probability density function of z Approximation via the following series:

[0063] in, It is the probability density function of the standard normal distribution. and They are The skewness and excess kurtosis are calculated from its third and fourth moments. n (z) is the nth order Hermitian polynomial. Using this formula, we obtain a polynomial about... The analytical, continuous probability density function accurately reflects the fluctuation characteristics of hot spot temperature caused by the randomness of charging load.

[0064] S4. Determination of the available charging load margin Based on the results of the preceding probabilistic analysis, the number of charging loads that can be connected to the transformer is finally determined.

[0065] S4.1 Definition of Probabilistic Security Constraints A risk-based assessment criterion is used, rather than a deterministic temperature limit. The safety constraint is defined as the transformer winding hot spot temperature at any given time during the assessment period. Exceeding the long-term operating temperature limit allowed by its insulation material The probability must be less than a very small threshold set by the grid operator based on its risk appetite. The mathematical expression is:

[0066] This probability can be derived from the probability density function of the standardized hotspot temperature obtained in step S3.3. To calculate using integration:

[0067] S4.2 Definition of Probabilistic Security Constraints Calculate the maximum number of new charging stations This ensures that the aforementioned probabilistic safety constraints still hold even after the addition of new charging stations. An incremental iterative algorithm is employed: (1) Initialization: Let the number of currently connected charging piles be . The iteration increment is i=1.

[0068] (2) Iterative loop: a. Set the total number of charging piles to be evaluated. ; b. Perform preceding calculations: Using the number of charging stations as input, complete the analytical calculations in steps S2 and S3 to obtain the probability density function of hotspot temperature at all times of day for that number of stations. . c. Validate constraints: Find the probability of a given day. Find the moment of maximum probability and determine whether this maximum probability satisfies the formula. Constraints.

[0069] d. Incremental iteration: Record the current... Let i = i + 1, return to step a for the next iteration, and continue to try to connect more charging piles.

[0070] e. Terminate iteration: If the safety constraints are not met, then... The number of charging stations has exceeded the safety margin, and the iteration process has been terminated.

[0071] (3) Determine the result: The maximum total number of securely accessible devices is .

[0072] (4) Output margin: The final margin that can be added is .

[0073] S4.3, Final Evaluation Result Output The final output is a clear, specific result that provides decision support.

[0074] Here is a calculation example: Transformer basic information: Rated active power 360 kW, current service life 15 years, historical average load: 60% Charging facilities to be connected: 30 new charging piles, each with a rated power of 7 kW.

[0075] Step 1: Calculate the dynamic power limit Quantifying health status: The overall health index (HI) of the transformer was calculated to be 63.4 by weighting the age score (62.5 points) and the historical load score (64 points).

[0076] Determining the reduction coefficient: Based on the health index, the dynamic reduction coefficient is calculated to be 0.895.

[0077] Determine the upper limit of power: Multiply the rated power by the health derating factor to obtain the upper limit of safe operating power for this transformer under the current conditions. P lim = 360 kW * 0.895 = 322.2 kW.

[0078] Step 2: Predict the total load for the next 24 hours Predicting base load: Based on the daily load curve of a typical residential area, predict the base load for the next 24 hours, excluding charging piles.

[0079] Predicting charging load: Based on the number of charging piles, rated power, and time-of-use charging simultaneity rate (charging demand is high at night, resulting in a high simultaneity rate), calculate the additional load that new charging facilities will bring.

[0080] Total load summed: The base load is added to the charging load to obtain the total load. P total This refers to the predicted total load curve for the transformer over the next 24 hours after it is connected to the charging pile.

[0081] Step 3: Calculate the available open capacity Calculate the margin: Subtract the predicted total load from the dynamic power limit to obtain the total margin.

[0082] Interpretation of results: If the result is positive, it means that there is remaining capacity during that period; if it is negative, it means that overload will occur during that period and the capacity will be insufficient.

[0083] Figure 2 The complete results of the example are presented intuitively, and the detailed analysis is as follows: Safety Boundary Analysis: The gray dashed line in the diagram represents the theoretical upper limit (rated power) of the transformer at 360 kW. The black dotted line represents our calculated actual safety upper limit (dynamic power limit) of 322.2 kW. This line takes into account the health condition of the equipment and is the true safety line that must be followed in the assessment.

[0084] Load composition and peak period analysis: The blue dashed line represents the base load, which peaks between 6 PM and 9 PM, but its peak value (approximately 288 kW) remains below the dynamic safety limit. When the additional charging load (red filled area) is added, the total load (red solid line) rises sharply after 6 PM, reaching a peak of approximately 414 kW around 7 PM. This far exceeds the black dynamic safety limit and even surpasses the rated power.

[0085] Available Capacity (Decision Basis) Analysis: The bar chart on the right is the final output of this assessment, clearly revealing key information for decision-making. Green Period (0:00-17:00): During these periods, the bar chart is above the 0 axis, indicating that the transformer has sufficient remaining capacity. Especially between 0:00 and 6:00 AM, the remaining capacity reaches over 145 kW, representing a golden window for guiding charging. Orange Period (18:00-23:00): During the evening peak hours, the bar chart turns orange and falls below the 0 axis, indicating insufficient transformer capacity and the potential for overload. At the most severe point at 19:00, the capacity deficit (i.e., overload) reached -91.8 kW.

[0086] Based on the above calculations and graphical analysis, the conclusions are as follows: Under the current assessment conditions, directly allowing 30 7kW charging piles to connect to the distribution transformer in a disorderly manner (i.e., users charging at any time) is not feasible. This connection method will cause continuous and severe overload to the transformer during the evening peak electricity consumption period (18:00-23:00), threatening not only equipment safety but also potentially affecting the normal electricity consumption of other users in the area. From 13:00 to 16:00, although the total load is not high, the P_limit curve itself contracts downward, significantly compressing the safety margin. This reveals another risk: if unexpected loads occur in the summer afternoon (such as temporary factory startup), the transformer may overload due to the low safety margin caused by ambient temperature. The orange and green bars representing the available capacity are calculated based on the time-varying dynamic power limit, showing the huge capacity gap at night and accurately reflecting that although there is still capacity in the afternoon (green bars), its margin is significantly smaller than in the morning and late at night.

[0087] Recommended measures: Implement an orderly charging strategy, using technological or electricity pricing mechanisms to guide most charging activity to the off-peak period between midnight and 6 AM. During this period, transformers have sufficient capacity to safely and economically meet the increased charging demand.

[0088] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for distribution transformer electric vehicle charging load dynamic capacity and risk assessment, characterized in that, The method comprises the following steps: S1, collecting multi-dimensional diagnostic data of the target transformer, constructing a comprehensive health index, and converting the comprehensive health index into a health derating coefficient through nonlinear mapping; S2, based on the historical load data of the transformer, using a time series decomposition method to obtain the basic load curve of the future period, and aggregating the load characteristic parameters of the batch charging piles based on the coincidence factor; S3, superimposing the deterministic basic load and the random charging load to obtain the total load of the transformer, using the health derating coefficient to correct the rated temperature rise parameter of the hot spot temperature response model, and then calculating the first four order central moments of the hot spot temperature through moment operation based on the statistical moments of the total load, and then using the Gram-Charlier series expansion method to construct the probability density function of the hot spot temperature; S4, based on the hot spot temperature probability distribution obtained in step S3, setting the probabilistic safety constraint of the hot spot temperature exceeding the limit; adjusting the number of charging piles through incremental iteration, and recalculating the hot spot temperature probability distribution under the new number to perform safety checking, determining the maximum number of accessible charging piles and outputting the additional margin.

2. The method of claim 1, wherein the method further comprises: The step S1 is specifically: S1.1, construction and quantification of the comprehensive health index: collecting multi-dimensional diagnostic data of the target transformer, the multi-dimensional diagnostic data including: dissolved gas analysis, insulation oil quality, insulation paper aging parameter, historical load and maintenance record, constructing a comprehensive health index HI value through standardized scoring and weighted summation of the multi-dimensional diagnostic data; S1.2, health derating coefficient: define the functional health derating coefficient of HI The piecewise nonlinear function is used to reflect the nonlinear characteristics of the deterioration speed of the health state. The segment points of the piecewise nonlinear function are based on the health level division of HI: 。 3. The method of claim 1, wherein the method further comprises: The time series decomposition method in step S2 is the Holt-Winters exponential smoothing method, which decomposes the historical daily load data into level, trend and seasonal components, and the calculation formula is: wherein, is the observed load value at time point t , , , are the level, trend and seasonal components at time point t , is the predicted value of the load value at time point t for the future h time steps, , , is the smoothing parameter, taking values in the range 0 to 1, m is the length of the seasonal cycle, h is the number of steps to forecast ahead, k = ( h 1) / h m , extrapolated after optimization of the smoothing parameter, to obtain the deterministic base load curve for the next 24 hours.

4. The method of claim 1, wherein the method further comprises: The calculation formula of the relationship between the coincidence factor and the number of charging piles in step S2 is: wherein , , are fitting parameters, is the number of charging piles.

5. The method of claim 4, wherein the method further comprises: The load characteristic parameters of the aggregated batch charging piles based on the coincidence factor are specifically: Aggregated peak power: Aggregated load mean: Aggregated load variance: wherein, is the rated power of a single charging station, denotes the probability that a single EV is in a charging state at time t.

6. The method of claim 1, wherein the method further comprises: The superposition of the deterministic basic load and the random charging load to obtain the total load of the transformer is specifically: The calculation formula of the total load of the transformer is: wherein, is the deterministic base load, is the stochastic charging load, the statistical moments of the total load are: mean , variance , skewness , kurtosis .

7. The method of claim 1, wherein the method further comprises: The hot spot temperature response model is specifically: Transformer hot spot temperature response model: Winding hot spot temperature rise equation: The final hot-spot temperature is: where is the ambient temperature, and are the temperature rise of the top oil relative to the ambient and the temperature rise of the hot-spot relative to the top oil, respectively, is the load factor, and are the corrected top oil temperature rise and hot-spot temperature rise at rated load, respectively, is the ratio of the rated load loss to the no-load loss, and are the thermal time constants of the oil and the winding, respectively, and are empirical exponents.

8. The method of claim 1, wherein the method further comprises: The calculation formula of the health derating coefficient for correcting the rated temperature rise parameter of the hot spot temperature response model is: wherein, Ttop is the top layer oil temperature rise at rated load, Thot is the hot spot temperature rise at rated load.

9. The method of claim 1, wherein, The probability density function of the hot spot temperature constructed by the Gram-Charlier series expansion method is specifically: The calculation formula of the standardization of the hot spot temperature is: wherein, is the standardized hotspot temperature, is the hotspot temperature is the statistical mean, is the hotspot temperature is the standard deviation; the probability density function of by the following series approximation: where is the probability density function of the standard normal distribution, and are the skewness and excess kurtosis of respectively, computed from its third and fourth moments, is the n th Hermite polynomial.

10. The method of claim 1, wherein the method further comprises: The S4 is specifically: S4.1, Set safety constraints: the probability that the transformer winding hot-spot temperature exceeds the long-term operating temperature limit allowed for the insulation material at any time within the assessment period is less than the risk threshold set by the grid operator which is mathematically expressed as: where the probability By integrating the probability density function of the normalized hot spot temperature is calculated by integrating the probability density function of the normalized hot spot temperature S4.2, calculate the maximum number of additional charging piles by incremental iteration, the specific steps are as follows: (1) initialization: set the current number of access charging piles as , iteration increment i = 1; (2) Iterative loop: a. Set the total number of charging piles to be evaluated currently ; b. Perform preliminary calculations: with As the number of charging piles input, the calculation process of step S2 and step S3 is performed to obtain the probability density function of the hotspot temperature at each time in the evaluation period under the number of ;​ c. Risk probability test: calculate the risk probability at each time in the evaluation time find the time with the maximum risk probability, determine whether the maximum risk probability meets the safety constraint condition, if yes, continue to step d, otherwise, execute step e; d. Incremental iteration: record current for acceptable number of charging stations; let i = i + 1, return to step a for next iteration; e. Terminate iteration: if the safety constraint condition is not satisfied, then the number of charging piles has exceeded the safety margin, the iteration process is terminated; (3) Determination result: the total number of maximum safe access is ; (4) Output margin: the final addable margin is ; S4.3, output the final evaluation result: output the final evaluation report containing the basic information of the transformer and the additional margin.

Images