Mobile energy storage based phase-splitting power network power calculation method and system

Abstract

This invention discloses a method and system for calculating phase-to-phase feeder power based on mobile energy storage, relating to the fields of distribution network voltage control and mobile energy storage dispatch technology. The method includes: acquiring the historical active power and phase voltage sequences of each phase on the low-voltage side of the distribution transformer, constructing a voltage-power mapping model for each phase; acquiring the real-time three-phase voltage at the grid connection point, determining voltage limits phase by phase, calculating the voltage deviation of the exceeding phase if voltage limits exist, and calculating the negative-sequence voltage imbalance if no voltage limits exist, identifying the problematic phase and solving for the minimum compensation voltage when the negative-sequence voltage imbalance exceeds the limit; substituting the voltage deviation or minimum compensation voltage into the mapping model of the corresponding phase to obtain the target feeder active power for that phase; performing engineering constraint verification on the target feeder active power and outputting the power that meets the engineering constraints. This invention achieves phase-by-phase and demand-based adaptive power compensation, enabling precise management of distribution network voltage exceeding limits and three-phase imbalance problems.

Description

Technical Field

[0001] This invention belongs to the field of distribution network voltage control and mobile energy storage dispatch technology, and particularly relates to a method and system for calculating phase-by-phase feeder power based on mobile energy storage. Background Technology

[0002] Two prominent power quality issues commonly exist on the low-voltage side of distribution networks: first, single-phase or multi-phase voltage exceeding limits caused by heavy overload, line voltage drop, and increased end-point power supply radius; second, three-phase voltage imbalance caused by unbalanced user phase connections and an increased proportion of single-phase electrical equipment. These problems directly affect power supply reliability and user experience, and are key challenges in distribution network operation and control.

[0003] To address the aforementioned issues, existing solutions primarily rely on two technical approaches: one is equipment control, including on-load tap changers, reactive power compensation devices, and three-phase imbalance automatic adjustment devices; the other is engineering upgrades, such as line capacity expansion and conductor replacement. These methods generally suffer from limitations such as long construction periods, high investment costs, and rigid location requirements, making them ill-suited to the time-varying characteristics of loads and the rapid fluctuations following the integration of distributed energy resources. More importantly, existing control strategies often employ consistent three-phase compensation or are based on experience-based tuning, failing to provide precise correction for specific phases and voltage deviations. This results in both redundant and insufficient compensation, making it difficult to guarantee effective control.

[0004] In recent years, mobile energy storage, especially autonomous driving energy storage vehicles, has provided a new technological path for distribution network voltage management due to its advantages such as mobility, rapid deployment, and proximity access. Theoretically, this approach can achieve on-demand operation and flexible relocation, effectively compensating for the spatial and temporal limitations of fixed equipment. However, current mobile energy storage applications remain at the level of single power command output for the entire vehicle, lacking a phase-by-phase quantitative calculation mechanism for grid connection voltage indicators. Specifically, existing methods cannot automatically complete the following key steps based on real-time voltage status: accurately identifying the problem type (voltage exceeding limits or three-phase imbalance), quantitatively determining the minimum compensation voltage to meet assessment requirements, and then calculating the active power command required for each phase. This results in a lack of precise matching between the control capabilities of mobile energy storage and actual voltage management needs, making it difficult to directly connect with assessment indicators such as voltage compliance rate and three-phase imbalance, leading to uncertainty in the compensation effect. Summary of the Invention

[0005] To address the aforementioned deficiencies in existing technologies, the present invention aims to provide a phase-by-phase power calculation method and system based on mobile energy storage, thereby solving the problem that existing mobile energy storage lacks a phase-by-phase power quantification calculation method and cannot achieve phase-by-phase and demand-based adaptive compensation based on real-time voltage indicators at the grid connection point, thus achieving precise management of voltage overruns and three-phase imbalances in the distribution network.

[0006] This invention solves the above-mentioned technical problems through the following technical solution: a method for calculating the power of a phase-fed grid based on mobile energy storage, comprising:

[0007] Obtain the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, and based on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, construct a voltage-power mapping model to characterize the mapping relationship between the active power of each phase and the corresponding phase voltage on the low-voltage side of the distribution transformer.

[0008] The system acquires the real-time three-phase voltage of the mobile energy storage grid connection point and determines whether there is a voltage over-limit based on the preset voltage qualified range for each phase. If there is a voltage over-limit, the system calculates the corresponding voltage deviation for each over-limit phase.

[0009] If there is no voltage limit violation, the negative sequence voltage imbalance is calculated based on the real-time three-phase voltage. If the negative sequence voltage imbalance exceeds the preset three-phase imbalance assessment index, the problem phase is determined, and the minimum compensation voltage that makes the negative sequence voltage imbalance meet the three-phase imbalance assessment index is solved for the determined problem phase.

[0010] Substitute the voltage deviation or the minimum compensation voltage into the voltage-power mapping model of the corresponding phase, and solve inversely to obtain the target active power of the feeder network for that phase;

[0011] The active power of the target feeder is checked against engineering constraints. If the engineering constraints are met, the power is output directly; otherwise, the power is reduced according to the limits of the engineering constraints before output.

[0012] This invention determines the phase of three-phase imbalance by judging voltage over-limits phase by phase, calculates the voltage deviation or minimum compensation voltage for the problematic phase, and substitutes it into the voltage-power mapping model of the corresponding phase to solve for the target active power of the feeder network for that phase, thus achieving precise correction of specific phases and specific deviations. Compared with the existing technology of implementing three-phase consistent compensation or empirical tuning, this invention avoids the problems of compensation redundancy or undercompensation, and significantly improves compensation accuracy and scheduling flexibility.

[0013] This invention uses voltage deviation and minimum compensation voltage as unified metrics, incorporating two types of problems—voltage exceeding limits and three-phase imbalance—into the same calculation framework: first, the problem type is determined; then, the compensation voltage required to achieve the voltage target or meet the imbalance requirements is calculated; finally, the active power is solved using the same mapping model. This approach provides clear and verifiable quantitative evidence for the control objective, avoiding the logical fragmentation and complex tuning issues caused by using different control strategies for the two types of problems in existing technologies.

[0014] This invention constructs a voltage-power mapping model based on the historical active power and historical phase voltage on the low-voltage side of the distribution transformer, without relying on complex equivalent grid parameters, line impedance, or load models. Compared to existing technologies that rely on tap changer voltage regulation, reactive power compensation, and other equipment that require on-site adjustments, this invention has good versatility and portability, can be quickly deployed in different distribution areas, and significantly reduces operation and maintenance costs and commissioning cycles.

[0015] After obtaining the target active power of the feeder network through inverse kinematics, this invention further verifies the engineering constraints: if the constraints are met, the power is output directly; otherwise, the power is reduced according to the limits of the engineering constraints before output. This mechanism ensures that the active power output by the energy storage vehicle does not exceed its physical limits, avoiding equipment overload or protection actions caused by theoretical calculations exceeding limits, and realizing closed-loop control between theoretical optimization and practical feasibility.

[0016] This invention integrates real-time voltage acquisition, problem identification, compensation calculation, power inverse solution, constraint verification, and output into a single process. Using grid connection point voltage measurement as the sole input, it automatically completes problem type identification, compensation calculation, and power command generation. Compared to the existing crude control method of mobile energy storage, which relies primarily on a single power command from the entire vehicle, this invention achieves rapid response and automatic adaptation to distribution network voltage quality issues, significantly improving the timeliness and automation level of control.

[0017] Furthermore, based on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, the voltage-power mapping model is constructed, including:

[0018] Data cleaning is performed on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer.

[0019] Based on the cleaned historical active power sequence and historical phase voltage sequence of each phase, voltage-power sample pairs of each phase are constructed and divided into training set and validation set according to a preset ratio.

[0020] The gradient boosting tree algorithm is adopted, with regression trees as base learners and the goal of minimizing the weighted loss function. Iterative training is performed on the training set, and an early stopping mechanism is used on the validation set to select the optimal model, thus obtaining the voltage-power mapping model for each phase.

[0021] This invention employs a gradient boosting tree algorithm with a regression tree as the base learner for iterative training, effectively fitting the nonlinear relationship between voltage and active power. Compared to linear models or simple regression methods, it exhibits higher fitting accuracy and generalization ability. By using a weighted loss function, reconstructed samples during data cleaning are assigned lower weights, effectively mitigating the impact of errors introduced by reconstructed samples on model fitting and improving model robustness. By dividing the training and validation sets and introducing an early stopping mechanism, overfitting is effectively prevented, ensuring that the obtained mapping model has good predictive performance on unseen data. The entire modeling process relies solely on historical data from the low-voltage side of the distribution transformer, without requiring additional information such as grid topology and line parameters, demonstrating good versatility and transferability, and can be quickly adapted to different distribution areas.

[0022] Furthermore, data cleaning is performed on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, including:

[0023] Set the detection window width and outlier discrimination coefficient. After removing null values, non-numerical values, placeholder anomalies, and duplicate timestamps, a sequence to be inspected is formed. ;in, Let be the voltage of the i-th phase on the low-voltage side of the distribution transformer at time t. Let i be the active power of the i-th phase on the low-voltage side of the distribution transformer at time t, where i takes the values ​​A, B, and C.

[0024] Based on the detection window width and outlier discrimination coefficient For voltage sequences respectively and power sequence Sliding window outlier detection: The detection window slides from the first data point in the voltage or power sequence, and the mean of the samples within the detection window is calculated at each step. with standard deviation For any data x within the detection window, if If so, the data will be marked as an outlier and removed.

[0025] Data reconstruction is performed on the null data segments resulting from the removal of outliers:

[0026] like Then, linear interpolation is performed using the endpoint data of the null data segment to fill all null values ​​in the null data segment;

[0027] like Then, the median of historical data that is in the same phase and at the same time point as the null value and belongs to other normal days is used to fill the null value, and the filled data is linearly interpolated and connected to the endpoint data of the null value data segment.

[0028] like If the null value data segment is deleted, all time periods corresponding to the null value data segment will be deleted directly, and the time periods corresponding to the null values ​​will no longer be retained;

[0029] Where L is the length of the null data segment; This is the lower limit threshold of the gap; This is the upper limit threshold of the gap.

[0030] This invention utilizes sliding window outlier detection to effectively identify and remove outlier samples in power data caused by measurement errors, communication anomalies, and other reasons, thus preventing abnormal data from interfering with subsequent model training. A hierarchical reconstruction strategy is employed for continuous missing data segments of varying lengths: short gaps are filled using linear interpolation to ensure data continuity and smoothness; medium-to-long gaps are filled with the median of historical data from the same phase, time, and other normal days, leveraging the periodicity of power load to make the reconstructed data more consistent with actual operating characteristics; long gaps are directly deleted based on the corresponding time, avoiding the introduction of unreliable filler values ​​due to excessive data loss. This hierarchical reconstruction strategy achieves a reasonable balance between data integrity and accuracy, providing high-quality training samples for subsequent mapping model training.

[0031] Furthermore, based on a preset voltage acceptable range, it is determined phase by phase whether there is a voltage over-limit, including:

[0032] Set the voltage acceptable range according to the distribution network voltage assessment indicators;

[0033] For each phase, if its real-time voltage exceeds the voltage acceptable range, then that phase is determined to have a voltage exceeding the limit.

[0034] For a phase that is determined to have a voltage exceeding the limit, the difference between its real-time voltage and the boundary value of the voltage acceptable range is calculated and used as the voltage deviation of that phase.

[0035] This invention sets a voltage acceptable range based on distribution network voltage assessment indicators and independently determines whether the real-time voltage exceeds this range for each phase. This accurately identifies the specific phase that exceeds the limit, providing a clear control target for subsequent phase-by-phase compensation. The voltage deviation is calculated using the difference between the real-time voltage and the boundary value of the acceptable range, which has a clear physical meaning and facilitates integration with subsequent mapping models, ensuring consistency between the compensation target and the power grid standards.

[0036] Furthermore, the negative sequence voltage imbalance is calculated based on the three-phase real-time voltage, including:

[0037] Based on the aforementioned three-phase real-time voltages, the positive-sequence voltage and negative-sequence voltage are calculated using the symmetrical component method:

[0038] ;

[0039] ;

[0040] in, , These are positive sequence voltage and negative sequence voltage, respectively. , , These are the real-time voltages of phases A, B, and C at the mobile energy storage grid connection point at time t; ;

[0041] Calculate the negative sequence voltage imbalance based on the positive sequence voltage and the negative sequence voltage:

[0042] ;

[0043] in, The negative sequence voltage imbalance at time t represents the mobile energy storage grid connection point. Calculate the modulus of the phasor.

[0044] This invention calculates positive-sequence and negative-sequence voltages using the symmetrical component method, and uses the percentage of the magnitudes of the negative-sequence and positive-sequence voltages as the negative-sequence voltage imbalance. The calculation results are accurate and have clear physical meaning, consistent with the assessment methods of relevant national standards (GB / T 15543-2008), facilitating direct integration with grid operation requirements. This calculation relies solely on the real-time three-phase voltage at the grid connection point, requiring no additional measuring equipment, thus simplifying engineering implementation.

[0045] Furthermore, if the negative sequence voltage imbalance exceeds a preset three-phase imbalance assessment index, then the problematic phase is determined, including:

[0046] If the negative sequence voltage imbalance exceeds the preset three-phase imbalance assessment index, then calculate the difference between the real-time voltage of each phase and the average real-time voltage of the three phases:

[0047] ;

[0048] ;

[0049] in, , , These are the real-time voltages of phases A, B, and C at the mobile energy storage grid connection point at time t; This represents the average real-time voltage of the three phases. The difference between the average real-time voltages of the three phases in the i-th phase is denoted as .

[0050] Will satisfy Furthermore, the phase with the smallest voltage amplitude is identified as the problematic phase; if the voltage difference between the phase with the second smallest voltage amplitude and the phase with the smallest voltage amplitude is less than a preset threshold, then both the phase with the smallest voltage amplitude and the phase with the second smallest voltage amplitude are identified as problematic phases.

[0051] This invention uses the difference between the real-time voltage of each phase and the three-phase average as the primary criterion. The phase with the smallest voltage amplitude and a difference less than 0 is identified as the problem phase. This accurately identifies the phase with the most severe voltage drop, which is typically the main contributing phase to the three-phase imbalance. Simultaneously, a threshold judgment is made based on the voltage difference between the phase with the second smallest voltage amplitude and the smallest phase to identify situations where both phases experience significant voltage drops, avoiding insufficient single-phase compensation. This determination method is logically clear, computationally simple, and can quickly locate the problem phase in real-time control, meeting the timeliness requirements of engineering applications.

[0052] Furthermore, the formula for calculating the minimum compensation voltage is as follows:

[0053] ;

[0054] in, This is the minimum compensation voltage for the i-th phase; The real-time voltage of the determined problem phase; This represents the average real-time voltage of the three phases.

[0055] This invention calculates the minimum compensation voltage by subtracting the real-time voltage amplitude of the problematic phase from the average three-phase voltage. This formula directly provides the compensation voltage required to raise the voltage of the problematic phase to the average three-phase level, with clear physical meaning. By targeting the average three-phase voltage, the three-phase voltages can be brought to equilibrium, thereby reducing the negative sequence voltage imbalance below the performance target. The calculation is simple, requires no iterative solution, and facilitates real-time control.

[0056] Furthermore, the active power of the target feeder network is subjected to engineering constraint verification, including:

[0057] Estimate the phase current based on the target active power of each phase and the target voltage of the corresponding phase:

[0058] ;

[0059] in, Let i be the current of phase i at time t. Let be the target active power of the feeder network in phase i at time t; Let be the target voltage of the i-th phase at time t; Let be the power factor of the i-th phase at time t;

[0060] If the phase current Greater than the upper limit of single-phase output current Then calculate the phase current limiting factor: ;

[0061] If the target feeder active power Exceeding the single-phase output power limit of mobile energy storage Then calculate the power limitation factor:

[0062] ;

[0063] Calculate the scaling factor: ;

[0064] If the scaling factor If the target active power of the feeder is satisfied, the target active power of the feeder will be directly output; otherwise, the target active power of the feeder will be scaled back according to the scaling factor, and the scaled power will be checked against engineering constraints until the engineering constraints are met.

[0065] This invention ensures that the output current of each phase of the mobile energy storage device does not exceed the device capacity through phase current estimation and single-phase output current upper limit verification, thus preventing equipment damage or protection activation due to overcurrent. Single-phase output power upper limit verification ensures that the output power of each phase of the energy storage vehicle does not exceed the rated capacity, guaranteeing safe operation. A scaling factor minimization method is used to comprehensively balance current and power constraints, ensuring that all constraints are simultaneously satisfied. A cyclic re-verification mechanism is employed to ensure that the power after retraction meets the constraints while avoiding insufficient power utilization due to excessive retraction in a single instance, achieving a balance between safety and economy.

[0066] Furthermore, the target voltage is determined based on the problem type:

[0067] If the problem is a voltage limit violation and the voltage exceeds the upper limit, then the target voltage is the upper limit of the voltage acceptable range, and the phase of the target voltage is consistent with the voltage phase of the corresponding phase.

[0068] If the problem is a voltage exceeding the limit and specifically a voltage exceeding the lower limit, then the target voltage is the lower limit of the acceptable voltage range, and the phase of the target voltage is consistent with the voltage phase of the corresponding phase.

[0069] If it is a three-phase imbalance problem, the target voltage is the average real-time voltage of the three phases, and the phase of the target voltage is the same as the phase of the voltage of the corresponding phase.

[0070] This invention sets different target voltages for different problem types, so that the determination of the target voltage is directly related to the problem type, forming a complete closed loop with the aforementioned judgment steps, ensuring that the compensation behavior is consistent with the governance goal, and at the same time, it is easy to connect with the voltage-power mapping model, simplifying the power inverse solution calculation.

[0071] Based on the same concept, the present invention also provides a phase-feed grid power calculation system based on mobile energy storage, including a memory, a processor, and a computer program or instructions stored in the memory, wherein the processor executes the computer program or instructions to implement the phase-feed grid power calculation method based on mobile energy storage as described above. Attached Figure Description

[0072] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only one embodiment of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0073] Figure 1 This is a flowchart of the phase-separated feeder power calculation method in an embodiment of the present invention;

[0074] Figure 2 This is a schematic diagram of the voltage-power mapping model fitting results in an embodiment of the present invention;

[0075] Figure 3 This is a schematic diagram illustrating the effect of energy storage vehicle feeder grid management during periods when the voltage exceeds the lower limit, as described in this embodiment of the invention.

[0076] Figure 4 This is a schematic diagram illustrating the phase separation and treatment effect of the three-phase unbalanced platform in an embodiment of the present invention. Detailed Implementation

[0077] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0079] Example 1

[0080] Reference Figure 1 This invention provides a method for calculating the power of a phase-fed grid based on mobile energy storage, specifically including the following steps:

[0081] Step S1: Obtain the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer.

[0082] The historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer are obtained from the measuring devices on the low-voltage side of the distribution transformer, such as distribution transformer terminal units (TTUs) or smart meters. The data acquisition cycle can be set according to the site conditions; in this embodiment, it is acquired once every 15 minutes. The time span of the historical data should cover the typical load cycle, for example, selecting historical data from the past 30 to 90 days to fully reflect the daily, weekly, and seasonal variations of the load.

[0083] The acquired historical data includes the voltage and active power sequences of phases A, B, and C, denoted as:

[0084] Voltage sequence: ,in t is the sampling time number. This refers to the voltage of the i-th phase on the low-voltage side of the distribution transformer at time t (i.e., the historical phase voltage, specifically the per-unit voltage value).

[0085] Active power sequence: ,in This refers to the active power (i.e., historical active power) of the i-th phase on the low-voltage side of the distribution transformer at time t.

[0086] The voltage value and active power value at each sampling time constitute a sample pair. The sample pairs are strictly aligned in time, meaning that the voltage and active power values ​​at the same moment correspond to the same operating conditions.

[0087] Step S2: Based on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, construct a voltage-power mapping model to characterize the mapping relationship between the active power of each phase and the corresponding phase voltage on the low-voltage side of the distribution transformer.

[0088] In a specific embodiment of the present invention, a voltage-power mapping model is constructed based on the historical active power sequence and the historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, specifically including:

[0089] Step S2.1: Perform data cleaning on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer.

[0090] Data cleaning of acquired historical data specifically includes the following sub-steps:

[0091] Step S2.11: Set the detection window width and outlier discrimination coefficient. After removing null values, non-numerical values, placeholder anomalies, and duplicate timestamps, a sequence to be inspected is formed. In this embodiment, the detection window width is set to 9, and the outlier discrimination coefficient is set to 3.

[0092] Null value: refers to the moment when the measuring device did not collect data, and the corresponding data field is empty;

[0093] Non-numeric: refers to data fields containing non-numeric characters, such as "NaN", "null", or garbled characters;

[0094] Abnormal placeholders: refers to the presence of obviously non-compliant placeholders in data fields, such as "9999" or "-9999";

[0095] Duplicate timestamps: These refer to records that exist at the same sampling time, and only one needs to be retained after deduplication.

[0096] After removing the above-mentioned abnormal data, the sequence to be inspected is formed. At each sampling time t, the voltage value and active power value are retained. If data is missing at a certain time due to the removal of outlier samples, the voltage or power value corresponding to that time will be temporarily empty and will be processed in the subsequent data reconstruction step.

[0097] Step S2.12: Based on the detection window width and outlier discrimination coefficient For voltage sequences respectively and power sequence Perform outlier detection using sliding window.

[0098] Specifically, starting from the first data point of the voltage or power sequence, the detection window slides in steps of size n, and the mean of the samples within the detection window is calculated each time. with standard deviation For any data x within the detection window, if If the data at time t is an outlier, it is marked as an outlier and removed, but the time after the removal is retained. For example, if the data at time t is an outlier, only the outlier data is removed, and time t is retained. In this embodiment, the step size n is 1.

[0099] Step S2.13: Reconstruct the data for the null data segment formed after removing outliers. The null data segment refers to a sequence consisting of at least one consecutive null value. Let the length of the null data segment be L, that is, the data corresponding to L consecutive time points are null values.

[0100] like If the null value is filled, linear interpolation reconstruction is used, that is, linear interpolation is performed using the endpoint data of the null value data segment to fill all null values ​​in the null value data segment.

[0101] like Then, the median of historical data that is in the same phase, at the same time point, and belongs to other normal days within the same week as the null value is used to fill the null value, and the filled data is linearly interpolated and connected to the endpoint data of the null value data segment.

[0102] For example, if the 10:00 AM data is missing on the 3rd day of the week (Tuesday), then the median of the 10:00 AM data from the 1st and 2nd days of the week (Sunday and Monday) and the 4th to 7th days (Wednesday to Saturday) is used to fill in the missing data for the 10:00 AM time on the 3rd day (Tuesday).

[0103] like If the null value data segment is deleted, all time periods corresponding to the null value data segment will be deleted directly, and the time periods corresponding to the null values ​​will no longer be retained.

[0104] For example, the data from 8:00 AM to 12:00 PM on the third day of the week (Tuesday) is missing, resulting in... Then delete the time period from 8:00 AM to 12:00 PM on the third day (Tuesday).

[0105] This is the lower limit threshold of the gap; This is the upper limit threshold for the gap. In this embodiment, the lower limit threshold for the gap is... The value is 3, which is the upper limit threshold of the gap. The value is 6.

[0106] Step S2.2: Based on the cleaned historical active power sequence and historical phase voltage sequence of each phase, construct voltage-power sample pairs for each phase and divide them into training set and validation set according to a preset ratio.

[0107] After data cleaning, voltage-power sample pairs for each phase are constructed based on the cleaned historical active power sequences and historical phase voltage sequences of each phase. The dataset was divided into training and training sets in an 8:2 ratio. and verification set .

[0108] Step S2.3: Using the gradient boosting tree algorithm, with the regression tree as the base learner and minimizing the weighted loss function as the objective, iterative training is performed on the training set, and an early stopping mechanism is used on the validation set to select the optimal model, thereby obtaining the voltage-power mapping model for each phase.

[0109] This embodiment uses a weighted loss function to construct the training objective, and solves for the model parameters by minimizing this training objective. Specifically, a voltage-power mapping relationship is defined for each phase; taking the i-th phase as an example:

[0110] (1)

[0111] in, The parameters to be solved in the model. This is a nonlinear mapping function obtained through training using the gradient boosting tree algorithm. In this embodiment, It is a model composed of a weighted sum of multiple regression trees, used to characterize the mapping relationship between the phase voltage and the active power of the feeder grid. Given a target voltage, the required target active power of the feeder grid for that phase can be obtained by inverse solving this model.

[0112] The training objective constructed using the weighted loss function can be expressed as:

[0113] (2)

[0114] in, For training objectives; The loss function; For the number of training iterations; Here are the weighting coefficients. In this embodiment, the loss function is the Huber loss function. Let the segmentation threshold of the Huber loss function be... (Taking the 80th percentile of the absolute residuals of the training set), the loss function is expressed as:

[0115] (3)

[0116] Constructing a base learner using a regression tree ,by Minimize as the training objective. As input, iteratively updated during training. Calculate the validation loss for each iteration on the validation set. If the current iteration number is less than or equal to the upper limit of iteration numbers (e.g., 100) and the validation loss reaches its minimum, or if the validation loss does not decrease for E consecutive iterations (e.g., 10), then stop model training and use the model at the time of stopping as the final voltage-power mapping model. .

[0117] Step S3: Obtain the three-phase real-time voltage of the mobile energy storage grid connection point, and determine whether the real-time voltage of each phase exceeds the preset voltage qualified range, thereby determining whether there is a voltage over-limit, and calculating the voltage deviation of the over-limit phase when the voltage over-limit is exceeded.

[0118] The voltage acceptable range is set according to the distribution network voltage assessment indicators. In this embodiment, the acceptable voltage range is set to a per-unit value [0.90, 1.07], corresponding to 90% to 107% of the rated voltage. Real-time voltages of each phase at the grid connection point of the mobile energy storage (e.g., an autonomous driving energy storage vehicle) are obtained. , .

[0119] For the transformer substations connected to the energy storage vehicle, a phase-by-phase determination is made to determine whether there is a voltage over-limit at the current moment. The determination rules are as follows:

[0120] (4)

[0121] in, This is an out-of-limit flag value. This indicates that at time t, the i-th phase has a voltage exceeding the limit. This indicates that the voltage of phase i is normal at time t. After the discrimination is completed, a label vector is formed. .

[0122] If the label vector If there are elements marked as 1, indicating a phase exceeding the voltage limit, then for each phase marked as 1, calculate the corresponding voltage deviation. :

[0123] (5)

[0124] If the label vector If the vector is all zeros, meaning that the real-time voltage of all phases is within the acceptable voltage range, it indicates that there is no voltage over-limit problem at the current moment. At this time, step S4 is executed to perform three-phase imbalance analysis.

[0125] Step S4: Calculate the negative sequence voltage imbalance based on the three-phase real-time voltage. If the negative sequence voltage imbalance exceeds the preset three-phase imbalance assessment index, then determine the problem phase and solve for the minimum compensation voltage that makes the negative sequence voltage imbalance meet the three-phase imbalance assessment index for the determined problem phase.

[0126] This step is executed after step S3 determines that there is no voltage over-limit issue, and is used to identify and quantify three-phase imbalance problems. Specifically, it includes the following sub-steps:

[0127] Step S4.1: Based on the three-phase real-time voltage, calculate the positive-sequence voltage and negative-sequence voltage using the symmetrical component method:

[0128] (6)

[0129] (7)

[0130] in, , These are positive sequence voltage and negative sequence voltage, respectively. , , These are the real-time voltages of phases A, B, and C at the mobile energy storage grid connection point at time t; ,Right now , Positive sequence voltage and negative sequence voltage All are phasors, containing amplitude and phase angle information.

[0131] Step S4.2: Calculate the negative sequence voltage imbalance based on the positive sequence voltage and the negative sequence voltage:

[0132] (8)

[0133] in, The negative sequence voltage imbalance at time t represents the mobile energy storage grid connection point. The magnitude of the phasor is calculated. The negative sequence voltage imbalance is expressed as a percentage, reflecting the proportion of the negative sequence component relative to the positive sequence component.

[0134] Referring to the short-term judgment criteria for three-phase imbalance in the national standard GB / T15543-2008 "Power Quality Three-Phase Voltage Imbalance", three-phase imbalance assessment indicators are set. In this embodiment, When the negative sequence voltage imbalance exceeds 4%, it is determined that there is a three-phase imbalance problem.

[0135] Step S4.3: Compare the negative sequence voltage imbalance with the three-phase imbalance assessment index to determine whether three-phase imbalance exists. The judgment rules are as follows:

[0136] like If so, it is determined that there is a three-phase imbalance, and step S4.4 is executed to determine the problematic phase;

[0137] like If the three-phase imbalance is not found, the mobile energy storage will not output power, i.e., the output power is 0.

[0138] Step S4.4: Determine the problem phase.

[0139] First, calculate the average real-time voltage of the three phases. :

[0140] (9)

[0141] Then calculate the difference between the real-time voltage of each phase and the average real-time voltage of the three phases. :

[0142] (10)

[0143] The rules for determining the phase of a problem are as follows:

[0144] Will satisfy Furthermore, the phase with the smallest voltage amplitude is identified as the problem phase. This phase is usually the one with the most severe voltage drop in the three phases and is also the main contributing phase to the three-phase imbalance.

[0145] If the voltage difference between the phase with the second smallest voltage amplitude and the phase with the smallest voltage amplitude is less than a preset threshold... The phase with the smallest voltage amplitude and the phase with the second smallest voltage amplitude are both identified as problematic phases. This rule is used to identify situations where both phases experience significant voltage drops, avoiding insufficient single-phase compensation. In this embodiment, a preset threshold is used. .

[0146] Step S4.5: Calculate the minimum compensation voltage for the determined problem phase. The specific calculation formula is as follows:

[0147] (11)

[0148] in, The real-time voltage of the determined problem phase. Formula (11) directly gives the compensation voltage required to raise the problem phase voltage to the level of the average three-phase voltage amplitude, with a clear physical meaning. With the three-phase average as the target, the three-phase voltage can be made to tend to balance, thereby reducing the negative sequence voltage imbalance to below the assessment index.

[0149] For each phase identified as a problem phase, the corresponding minimum compensation voltage is calculated according to the above formula (11). If only one phase is identified as a problem phase, the minimum compensation voltage of that phase is calculated only; if both phases are identified as problem phases, the minimum compensation voltage of each phase is calculated separately.

[0150] Step S5: Substitute the voltage deviation or minimum compensation voltage into the voltage-power mapping model of the corresponding phase, and solve the inverse to obtain the target active power of the feeder network for that phase.

[0151] Based on the determination results of steps S3 and S4, this step converts the calculated voltage deviation or minimum compensation voltage into the target active power of the feeder grid for each phase, which is then used as the power command output by the energy storage vehicle.

[0152] Depending on the type of problem, this step is divided into two scenarios for separate handling:

[0153] Scenario 1: Voltage exceeding limit scenario.

[0154] When step S3 determines that there is a voltage over-limit, the corresponding target feeder active power is calculated for each over-limit phase.

[0155] For the i-th phase that is determined to have a voltage over-limit, its voltage deviation is... Substitute into the voltage-power mapping model trained for this phase. The inverse solution yields the target active power of the feeder network for that phase. :

[0156] (12)

[0157] in, This refers to the target voltage amplitude after compensation required to restore the voltage to the acceptable range.

[0158] For phases where there is no voltage over-limit, there is no need to calculate the active power of the target feeder, i.e., no compensation power is output.

[0159] Scenario 2: Three-phase imbalance scenario.

[0160] When step S4 determines that a three-phase imbalance exists, the corresponding target feeder active power is calculated for each identified problematic phase. For the i-th phase identified as a problematic phase, its minimum compensation voltage is calculated. Substitute into the voltage-power mapping model trained for this phase. The inverse solution yields the target active power of the feeder network for that phase. :

[0161] (13)

[0162] in, This refers to the target voltage amplitude required to bring the three-phase voltages to a state of equilibrium.

[0163] For other phases not identified as problematic phases, there is no need to calculate the active power of the target feeder, i.e., no compensation power is output.

[0164] The target active power of the feeder network for each phase is obtained through the above calculations. ,in:

[0165] For voltage over-limit scenarios: only the over-limit phase has a non-zero value. The non-limited phase is 0;

[0166] Step S6: Perform engineering constraint verification on the target feeder active power. If the engineering constraints are met, output directly; otherwise, output after power reduction according to the limits of the engineering constraints.

[0167] This step involves calculating the target active power of each phase of the feeder obtained in step S5. Perform engineering constraint verification to ensure that the active power output of the energy storage vehicle does not exceed its physical limits. Engineering constraints include at least the upper limit of single-phase output current and the upper limit of single-phase output power of the energy storage vehicle. Specifically, this includes the following sub-steps:

[0168] Step S6.1: Estimate the phase current based on the target active power of each phase and the target voltage of the corresponding phase:

[0169] (14)

[0170] in, Let i be the current of phase i at time t. Let be the target active power of the feeder network in phase i at time t; Let be the target voltage of the i-th phase at time t; Let be the power factor of the i-th phase at time t. In this embodiment, .

[0171] The target voltage is determined based on the problem type, expressed in phasor form, and is associated with the power factor angle.

[0172] If the problem involves voltage exceeding the limit, specifically exceeding the upper limit, then the target voltage should be the upper limit of the acceptable voltage range, and the phase of the target voltage should be consistent with the phase of the corresponding phase. Power factor angle It is the difference between the phase of the feed current (calculated from the feed power and the voltage of the corresponding phase) and the phase of the voltage of the corresponding phase;

[0173] If the problem is a voltage exceeding the limit, specifically below the lower limit, then the target voltage should be the lower limit of the acceptable voltage range, and its phase should be consistent with the current power factor angle. Power factor angle It is the difference between the phase of the feed current and the phase of the corresponding phase voltage;

[0174] If it is a three-phase imbalance problem, the target voltage is the average of the three-phase voltage amplitudes, and the phase is consistent with the current power factor angle, that is... .

[0175] in, The power factor angle of the i-th phase at time t is determined by the current operating conditions.

[0176] Step S6.2: Set the upper limit of single-phase output current This value is determined by the hardware capacity of the energy storage vehicle. For each phase, if the phase current... Greater than the upper limit of single-phase output current Then calculate the phase current limiting factor:

[0177] (15)

[0178] If the phase current does not exceed the upper limit of the single-phase output current If the corresponding phase current limiting factor is 1 (or does not participate in the minimum value constraint in the scaling factor calculation), then the corresponding phase current limiting factor is 1.

[0179] Step S6.3: Set the upper limit of single-phase output power of the energy storage vehicle This value is determined by the rated capacity of the energy storage vehicle. For each phase, if the target active power of the feeder grid for that phase... Greater than the upper limit of single-phase output power of mobile energy storage Then calculate the power limitation factor for that phase:

[0180] (16)

[0181] If the target grid active power does not exceed the single-phase output power limit. If the power limitation factor is 1, then the corresponding power limitation factor is 1 (or it is not included in the minimum value constraint in the scaling factor calculation).

[0182] Step S6.4: Calculate the scaling factor :

[0183] (17)

[0184] like This indicates that all engineering constraints have been met, and the target active power of each phase feeder calculated in step S5 is directly output. .

[0185] like This indicates that engineering constraints are not met, and power needs to be scaled back according to the scaling factor. To prevent the power from remaining close to the upper limit after scaling back, leading to cyclic oscillations, the scaling back is performed at 90% of the scaling factor.

[0186] (18)

[0187] And recalculate the output power:

[0188] (19)

[0189] Step S6.5: Cyclic re-verification.

[0190] Based on the recalculated output power Repeat steps S6.1 to S6.4 for iterative verification until all engineering constraints are met (i.e., ...). At this point, the final power command is output.

[0191] The aforementioned cyclic re-verification mechanism ensures that the power after retraction meets the equipment safety constraints, while avoiding insufficient power utilization due to excessive retraction in a single instance, thus achieving a balance between safety and economy.

[0192] Figure 2 The voltage-power curve fitting results based on the method of this invention are shown, where the horizontal axis represents the active power on the low-voltage side of the distribution transformer, and the vertical axis represents the per-unit voltage value on the low-voltage side of the distribution transformer in the area. Figure 2 It can be seen that as the active power on the low-voltage side of the distribution transformer increases, the historical voltage sample generally shows a downward trend. The fitted voltage-power curve is basically consistent with the sample distribution trend, reflecting the change law of the distribution transformer voltage in the distribution area gradually approaching the lower limit of the voltage qualification range from the rated level. The above results show that the method of the present invention can better characterize the mapping relationship between the active power on the low-voltage side of the distribution transformer and the distribution transformer voltage in the distribution area, providing a basis for solving the target feeder active power in the subsequent process.

[0193] Figure 3The study demonstrates the changes in transformer voltage before and after the energy storage vehicle feeds into the grid during the period when the voltage exceeds the lower limit. The period from 18:45 to 20:15 represents the voltage exceeding the lower limit. During this period, if the energy storage vehicle does not feed into the grid, the transformer voltage is significantly lower than the lower limit of the acceptable voltage range (0.90 pu). However, after the energy storage vehicle outputs the active power calculated using the method of this invention, the transformer voltage is generally higher than before feeding into the grid, showing a significant increase during the period exceeding the limit. These results indicate that the active power of the energy storage vehicle feeding into the grid calculated by the method of this invention during the voltage exceeding the limit has a good correlation with the transformer voltage increase process, verifying the effectiveness of this method in solving for active power in the grid.

[0194] Figure 4 This demonstrates the changes in transformer voltage and imbalance degree in the distribution area before and after the energy storage vehicle feeds into the grid during the three-phase voltage imbalance period, with 18:00-21:30 being the imbalance period. Figure 4 It can be seen that during this period, the voltage of phase A is at a relatively low level before the energy storage vehicle feeds into the grid, and is generally higher after feeding into the grid than before. Simultaneously, the voltage imbalance is higher than the 2% three-phase voltage imbalance margin before feeding into the grid, and is generally lower than this margin after feeding into the grid. These results indicate that the method of this invention can calculate the corresponding active power of the phase-specific grid feed based on the voltage changes and imbalance constraints of the problematic phase, providing a basis for phase-specific grid feed control of energy storage vehicles.

[0195] Example 2

[0196] This invention also provides a phase-feed grid power calculation system based on mobile energy storage. The system includes a memory, a processor, and a computer program or instructions stored in the memory. The processor executes the computer program or instructions to implement the phase-feed grid power calculation method based on mobile energy storage in this invention.

[0197] Although not shown, the system includes a processor that can perform various appropriate operations and processes based on programs and / or data stored in read-only memory (ROM) or loaded from a storage portion into random access memory (RAM). The processor can be a multi-core processor or may contain multiple processors. In some embodiments, the processor may include a general-purpose main processor and one or more specialized coprocessors, such as a central processing unit, graphics processing unit (GPU), neural network processor (NPU), digital signal processor (DSP), etc. Various programs and data required for device operation are also stored in the RAM. The processor, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0198] The processor and memory described above are used together to execute programs / instructions stored in the memory. When the program / instructions are executed by the computer, they can implement the methods, steps, or functions described in the above embodiments.

[0199] The above description only discloses specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or modifications that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A mobile energy storage based phase-splitting power calculation method for a distribution network, characterized in that, The method includes: Obtain the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, and based on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, construct a voltage-power mapping model to characterize the mapping relationship between the active power of each phase and the corresponding phase voltage on the low-voltage side of the distribution transformer. The system acquires the real-time three-phase voltage of the mobile energy storage grid connection point and determines whether there is a voltage over-limit based on the preset voltage qualified range for each phase. If there is a voltage over-limit, the system calculates the corresponding voltage deviation for each over-limit phase. If there is no voltage limit violation, the negative sequence voltage imbalance is calculated based on the real-time three-phase voltage. If the negative sequence voltage imbalance exceeds the preset three-phase imbalance assessment index, the problem phase is determined, and the minimum compensation voltage that makes the negative sequence voltage imbalance meet the three-phase imbalance assessment index is solved for the determined problem phase. Substitute the voltage deviation or the minimum compensation voltage into the voltage-power mapping model of the corresponding phase, and solve inversely to obtain the target active power of the feeder network for that phase; The active power of the target feeder is checked against engineering constraints. If the engineering constraints are met, the power is output directly; otherwise, the power is reduced according to the limits of the engineering constraints before output.

2. The mobile energy storage based split phase network power calculation method of claim 1, wherein, Based on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, the voltage-power mapping model is constructed, including: Data cleaning is performed on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer. Based on the cleaned historical active power sequence and historical phase voltage sequence of each phase, voltage-power sample pairs of each phase are constructed and divided into training set and validation set according to a preset ratio. The gradient boosting tree algorithm is adopted, with regression trees as base learners and the goal of minimizing the weighted loss function. Iterative training is performed on the training set, and an early stopping mechanism is used on the validation set to select the optimal model, thus obtaining the voltage-power mapping model for each phase.

3. The method for calculating the power of a phase-fed grid based on mobile energy storage according to claim 2, characterized in that, Data cleaning is performed on the historical active power sequence and historical phase voltage sequence of each phase on the low-voltage side of the distribution transformer, including: Set the detection window width and outlier discrimination coefficient. After removing null values, non-numerical values, placeholder anomalies, and duplicate timestamps, a sequence to be inspected is formed. ;in, Let be the voltage of the i-th phase on the low-voltage side of the distribution transformer at time t. Let i be the active power of the i-th phase on the low-voltage side of the distribution transformer at time t, where i takes the values ​​A, B, and C. Based on the detection window width and outlier discrimination coefficient For voltage sequences respectively and power sequence Sliding window outlier detection: The detection window slides from the first data point in the voltage or power sequence, and the mean of the samples within the detection window is calculated at each step. with standard deviation For any data x within the detection window, if If so, the data will be marked as an outlier and removed. Data reconstruction is performed on the null data segments resulting from the removal of outliers: like Then, linear interpolation is performed using the endpoint data of the null data segment to fill all null values ​​in the null data segment; like Then, the median of historical data that is in the same phase and at the same time point as the null value and belongs to other normal days is used to fill the null value, and the filled data is linearly interpolated and connected to the endpoint data of the null value data segment. like If the null value data segment is deleted, all time periods corresponding to the null value data segment will be deleted directly, and the time periods corresponding to the null values ​​will no longer be retained; Where L is the length of the null data segment; This is the lower limit threshold of the gap; This is the upper limit threshold of the gap.

4. The method for calculating the power of a phase-fed grid based on mobile energy storage according to claim 1, characterized in that, Based on a preset voltage acceptable range, determine phase by phase whether there is a voltage over-limit, including: Set the voltage acceptable range according to the distribution network voltage assessment indicators; For each phase, if its real-time voltage exceeds the voltage acceptable range, then that phase is determined to have a voltage exceeding the limit. For a phase that is determined to have a voltage exceeding the limit, the difference between its real-time voltage and the boundary value of the voltage acceptable range is calculated and used as the voltage deviation of that phase.

5. The method for calculating the power of a phase-fed grid based on mobile energy storage according to claim 1, characterized in that, The negative sequence voltage imbalance is calculated based on the three-phase real-time voltage, including: Based on the aforementioned three-phase real-time voltages, the positive-sequence voltage and negative-sequence voltage are calculated using the symmetrical component method: ； ； in, , These are positive sequence voltage and negative sequence voltage, respectively. , , These are the real-time voltages of phases A, B, and C at the mobile energy storage grid connection point at time t; ; Calculate the negative sequence voltage imbalance based on the positive sequence voltage and the negative sequence voltage: ； in, The negative sequence voltage imbalance at time t represents the mobile energy storage grid connection point. Calculate the modulus of the phasor.

6. The method for calculating the power of a phase-fed grid based on mobile energy storage according to claim 1, characterized in that, If the negative sequence voltage imbalance exceeds the preset three-phase imbalance assessment index, then the problematic phase is determined, including: If the negative sequence voltage imbalance exceeds the preset three-phase imbalance assessment index, then calculate the difference between the real-time voltage of each phase and the average real-time voltage of the three phases: ； ； in, , , These are the real-time voltages of phases A, B, and C at the mobile energy storage grid connection point at time t; This represents the average real-time voltage of the three phases. The difference between the average real-time voltages of the three phases in the i-th phase is denoted as . Will satisfy Furthermore, the phase with the smallest voltage amplitude is identified as the problematic phase; if the voltage difference between the phase with the second smallest voltage amplitude and the phase with the smallest voltage amplitude is less than a preset threshold, then both the phase with the smallest voltage amplitude and the phase with the second smallest voltage amplitude are identified as problematic phases.

7. The method for calculating the power of a phase-fed grid based on mobile energy storage according to claim 6, characterized in that, The formula for calculating the minimum compensation voltage is as follows: ； in, This is the minimum compensation voltage for the i-th phase; The real-time voltage of the determined problem phase; This represents the average real-time voltage of the three phases.

8. The method for calculating the power of a phase-fed grid based on mobile energy storage according to any one of claims 1 to 7, characterized in that, The engineering constraint verification of the target feeder active power includes: Estimate the phase current based on the target active power of each phase and the target voltage of the corresponding phase: ； in, Let i be the current of phase i at time t. Let be the target active power of the feeder network in phase i at time t; Let be the target voltage of the i-th phase at time t; Let be the power factor of the i-th phase at time t; If the phase current Greater than the upper limit of single-phase output current Then calculate the phase current limiting factor: ; If the target feeder has active power Exceeding the single-phase output power limit of mobile energy storage Then calculate the power limitation factor: ； Calculate the scaling factor: ; If the scaling factor If the target active power of the feeder is satisfied, the target active power of the feeder will be directly output; otherwise, the target active power of the feeder will be scaled back according to the scaling factor, and the scaled power will be checked against engineering constraints until the engineering constraints are met.

9. The method for calculating the power of a phase-fed grid based on mobile energy storage according to claim 8, characterized in that, The target voltage is determined based on the problem type: If the problem is a voltage limit violation and the voltage exceeds the upper limit, then the target voltage is the upper limit of the voltage acceptable range, and the phase of the target voltage is consistent with the voltage phase of the corresponding phase. If the problem is a voltage exceeding the limit and specifically a voltage exceeding the lower limit, then the target voltage is the lower limit of the acceptable voltage range, and the phase of the target voltage is consistent with the voltage phase of the corresponding phase. If it is a three-phase imbalance problem, the target voltage is the average real-time voltage of the three phases, and the phase of the target voltage is consistent with the phase of the voltage of the corresponding phase.

10. A phase-fed grid power calculation system based on mobile energy storage, comprising a memory, a processor, and a computer program or instructions stored in the memory, characterized in that, The processor executes the computer program or instructions to implement the phase-feed grid power calculation method based on mobile energy storage as described in any one of claims 1 to 9.

Images