Multi-factor coupling energy storage power station real-time power compensation method and device

Abstract

The embodiment of the invention provides a multi-factor coupling energy storage power station real-time power compensation method and device. The method comprises the following steps: constructing a multivariable coupling model; performing parameter initialization setting on the multivariable coupling model; collecting and preprocessing multi-dimensional data; real-time optimization of multivariable coefficients of the multivariable coupling model is realized through gain matrix calculation and iterative updating of coefficient vectors, and updated multivariable coefficients are obtained; and closed-loop power control is realized based on the updated multivariable coefficient. According to the scheme of the invention, the requirements of GB / T34120-2023 on primary frequency modulation response time (less than or equal to 100ms) and adjustment precision can be met; the problems that a dynamic constraint mechanism of SOC and temperature and power output is lacked, and safety risks such as over-charging, over-discharging and overheat power impact exist are solved.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a real-time power compensation method for a multi-factor coupled energy storage power station, a real-time power compensation device for a multi-factor coupled energy storage power station, a computer device, and a computer-readable storage medium. Background Technology

[0002] With the increasing penetration of new energy sources (wind power, photovoltaics, etc.) in the power system, the power grid is placing more stringent requirements on the power response speed, compensation accuracy, and operational safety of energy storage power stations. According to GB / T34120-2023, when a power storage converter (PCS) participates in primary frequency regulation, its response time must be ≤100ms, its regulation time ≤300ms, and it must have four-quadrant power output capability. The closest existing technology to this invention is the traditional least-squares power compensation method based on a single variable. Typically, this method uses only single factors such as load current and new energy output as variables to construct a quadratic function. While existing technologies can be used to fit and compensate for energy storage power, they suffer from several drawbacks: Limited model dimensionality: Traditional methods often employ a single variable (such as load current) to construct a quadratic function fitting model, reflecting only the impact of a single factor on power and neglecting the coupling effect of core state parameters of the energy storage system. Dynamic characteristic studies of lithium-ion batteries show that parameters such as state of charge (SOC) and temperature (T) significantly affect power output characteristics, and their internal resistance and capacitance exhibit significant dynamic changes under different operating conditions. Poor adaptability to time-varying conditions: Conventional least squares methods use fixed-coefficient models, which cannot adapt to the time-varying characteristics of energy storage systems. During charging and discharging, the battery's internal parameters gradually change with the number of cycles; for example, resistances R1 and R2 gradually increase over time, while capacitance C1 changes rapidly. Fixed models struggle to track these changes, leading to deterioration in compensation accuracy after long-term operation. Data saturation: Traditional recursive least squares (RLS) suffers from "data saturation" during parameter identification, where historical data accumulation prevents the identification results from reflecting new data characteristics, hindering timely responses to rapid power fluctuations in the grid. Summary of the Invention

[0003] In view of the above problems, embodiments of the present invention are proposed to provide a real-time power compensation method for a multi-factor coupled energy storage power station, a real-time power compensation device for a multi-factor coupled energy storage power station, a computer device, and a computer-readable storage medium to overcome or at least partially solve the above problems.

[0004] To address the aforementioned problems, this invention discloses a real-time power compensation method for a multi-factor coupled energy storage power station, the method comprising:

[0005] Construct a multivariate coupled model;

[0006] Perform parameter initialization settings for multivariable coupled models;

[0007] Multi-dimensional data collection and preprocessing;

[0008] By calculating the gain matrix and iteratively updating the coefficient vector, the multivariable coefficients of the multivariable coupled model are optimized in real time, and the updated multivariable coefficients are obtained.

[0009] Closed-loop power control is achieved based on the updated multivariable coefficients.

[0010] Preferably, the construction of the multivariate coupled model includes:

[0011] We obtained the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0012] A multivariate coupled model is constructed using the aforementioned basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0013] Preferably, the parameter initialization settings for the multivariable coupled model include:

[0014] Set the initial coefficient vector as a multivariate coefficient vector;

[0015] Set the covariance matrix initialization and forgetting factor.

[0016] Preferably, the multi-dimensional data acquisition and preprocessing includes:

[0017] A distributed data acquisition architecture is adopted to achieve synchronous acquisition of multi-dimensional parameters. The acquisition device is configured through current sensor, temperature sensor, BMS interface and power sensor. The sampling period is 100ms. All sensor data are synchronized by timestamp to construct multivariate design vector.

[0018] Preferably, the method of calculating the gain matrix and iteratively updating the coefficient vector to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, and obtaining the updated multivariable coefficients, includes:

[0019] The weights of dynamically adjusted historical data are determined as the forgetting factor;

[0020] Through the first The covariance matrix at time t, the first The gain matrix is ​​calculated by using a multivariate design vector and forgetting factor at each time step.

[0021] Through the first The measured power at each time step, combined with the gain matrix, is used to incrementally correct the current parameter vector, and the coefficient vector is iteratively updated to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, resulting in the updated multivariable coefficients and covariance matrix. Adjustments are made synchronously as the multivariate coefficients are updated.

[0022] This invention discloses a real-time power compensation device for a multi-factor coupled energy storage power station, the device comprising:

[0023] Modules for building multivariate coupled models;

[0024] The initialization settings module is used to initialize parameters for multivariable coupled models.

[0025] The data acquisition module is used for multi-dimensional data acquisition and preprocessing.

[0026] The update module is used to optimize the multivariate coefficients of a multivariate coupled model in real time by calculating the gain matrix and iteratively updating the coefficient vector, so as to obtain the updated multivariate coefficients.

[0027] The control module is used to implement closed-loop power control based on the updated multivariable coefficients.

[0028] Preferably, the building module includes:

[0029] The acquisition submodule is used to acquire the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0030] A submodule is constructed to build a multivariate coupled model using the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0031] Preferably, the initialization setting module includes:

[0032] The first setting submodule is used to set the coefficient vector at the initial time as a multivariable coefficient vector;

[0033] The second settings submodule is used to set the covariance matrix initialization and forgetting factor.

[0034] This invention discloses a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described multi-factor coupled real-time power compensation method for energy storage power stations.

[0035] This invention discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described multi-factor coupled real-time power compensation method for energy storage power stations.

[0036] The embodiments of the present invention have the following advantages:

[0037] In this embodiment of the invention, traditional univariate fitting models cannot reflect the coupled influence of multiple factors such as SOC, temperature, and power fluctuation type on the compensation power, resulting in low compensation accuracy, especially with errors exceeding ±5% under edge conditions. The fixed coefficient model of this application can adapt to time-varying characteristics such as battery aging and temperature drift, and its accuracy continues to improve after long-term operation. Conventional RLS algorithms suffer from data saturation problems and cannot effectively track rapid power fluctuations in the power grid. The solution of this application can meet the requirements of GB / T34120-2023 for primary frequency regulation response time (≤100ms) and regulation accuracy. It solves the problem of lacking a dynamic constraint mechanism for SOC and temperature and power output, which poses safety risks such as overcharging, over-discharging, and overheating power surges. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a schematic diagram of an embodiment of a multi-factor coupled real-time power compensation method for an energy storage power station according to an embodiment of the present invention;

[0040] Figure 2 This is a flowchart illustrating an embodiment of a real-time power compensation method for a multi-factor coupled energy storage power station according to an embodiment of the present invention.

[0041] Figure 3 This is a schematic diagram of a real-time power compensation device for a multi-factor coupled energy storage power station according to an embodiment of the present invention.

[0042] Figure 4 This is an internal structural diagram of a computer device according to one embodiment. Detailed Implementation

[0043] To make the technical problems, technical solutions, and beneficial effects solved by the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0044] Reference Figure 1 The diagram illustrates an embodiment of a real-time power compensation method for a multi-factor coupled energy storage power station according to the present invention, which may specifically include the following steps:

[0045] Step S101: Construct a multivariate coupled model;

[0046] Step S102: Initialize the parameters for the multivariable coupled model;

[0047] Step S103: Multi-dimensional data acquisition and preprocessing;

[0048] Step S104: Real-time optimization of the multivariable coefficients of the multivariable coupled model is achieved by calculating the gain matrix and iteratively updating the coefficient vector, resulting in updated multivariable coefficients.

[0049] Step S105: Implement closed-loop power control based on the updated multivariable coefficients.

[0050] Furthermore, the construction of the multivariate coupled model includes:

[0051] We obtained the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0052] A multivariate coupled model is constructed using the aforementioned basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0053] In one specific embodiment, the parameter initialization setting for the multivariable coupled model includes:

[0054] Set the initial coefficient vector as a multivariate coefficient vector;

[0055] Set the covariance matrix initialization and forgetting factor.

[0056] It should be noted that the multi-dimensional data acquisition and preprocessing includes:

[0057] A distributed data acquisition architecture is adopted to achieve synchronous acquisition of multi-dimensional parameters. The acquisition device is configured through current sensor, temperature sensor, BMS interface and power sensor. The sampling period is 100ms. All sensor data are synchronized by timestamp to construct multivariate design vector.

[0058] In this embodiment of the invention, the method of real-time optimization of the multivariable coefficients of a multivariable coupled model through gain matrix calculation and iterative updating of coefficient vectors to obtain updated multivariable coefficients includes:

[0059] The weights of dynamically adjusted historical data are determined as the forgetting factor;

[0060] Through the first The covariance matrix at time t, the first The gain matrix is ​​calculated by using a multivariate design vector and forgetting factor at each time step.

[0061] Through the first The measured power at each time step, combined with the gain matrix, is used to incrementally correct the current parameter vector, and the coefficient vector is iteratively updated to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, resulting in the updated multivariable coefficients and covariance matrix. Adjustments are made synchronously as the multivariate coefficients are updated.

[0062] In a specific example of an embodiment of the present invention, the process steps of the present invention are as follows: Figure 2 As shown,

[0063] 1. Construction of a multivariate coupled model;

[0064] Breaking through the limitations of traditional univariate models, a power compensation model integrating multiple factors is constructed: ;

[0065] in:

[0066] The basic power impact factor is the load current or new energy output collected in real time at the grid connection point of the energy storage power station, and the unit is A;

[0067] : Battery state of charge, with a value ranging from 0 to 1, obtained in real time through the battery management system (BMS);

[0068] Battery temperature, in °C, is collected by the temperature sensor built into the battery cell.

[0069] Power fluctuation type identifier This indicates smooth fluctuations (power change rate ≤ 5% / ms). This indicates impact fluctuations (power change rate > 5% / ms);

[0070] The multivariate fitting coefficients to be determined are solved in real time using the recursive least squares method with a forgetting factor.

[0071] 2. Model parameter initialization;

[0072] Parameter initialization settings for multivariable coupled models:

[0073] Multivariable coefficient vector initialization: The coefficient vector at the initial time (k=0) is set as follows:

[0074]

[0075] The parameters are set based on factory calibration values ​​for typical energy storage conditions and can be adjusted according to the actual battery type.

[0076] Covariance matrix initialization: The initial value of the covariance matrix is ​​set to... ,in , The initial model is a 6×6 identity matrix, ensuring that it is highly sensitive to the first batch of data and achieves rapid convergence.

[0077] The covariance matrix in this invention The computational objects are the six coefficients to be fitted in the multivariate coupled model. (Corresponding model formula) Therefore, the matrix dimension is 6×6. The core function of this matrix is ​​to quantify the "uncertainty of the estimated values ​​of each coefficient to be fitted": the elements on the diagonal of the matrix correspond to... The individual estimated variances (the larger the value, the lower the reliability of the coefficient), and the estimated covariance between any two coefficients corresponding to off-diagonal elements (reflecting the uncertainty of the correlation between the estimation results of different coefficients). Specifically, this manifests as follows:

[0078] ;

[0079] Use during initialization The core reason for the calculation method lies in: the initial time ( The coefficients to be fitted For empirical calibration values, high uncertainty needs to be introduced through "large numerical variance" to ensure the accuracy of the initial batch of multi-dimensional data (current). SOC, temperature (T), power fluctuation type (etc.) can quickly correct coefficients and shorten the model convergence time.

[0080] Forgetting Factor Setting: Forgetting Factor The value is set to 0.97. This value is based on extensive experimental verification and can effectively reduce the influence of historical data, avoid data saturation, and ensure model stability. When power fluctuations are severe (e.g., δ=1), it can be dynamically adjusted to 0.95 to speed up the response; under stable operating conditions (δ=0), it can be adjusted to 0.99 to enhance stability.

[0081] 3. Multi-dimensional data collection and preprocessing;

[0082] A distributed data acquisition architecture is adopted to achieve synchronous acquisition of multi-dimensional parameters;

[0083] Data acquisition equipment configuration: Current sensor: Hall effect current sensor, measurement range 0~1000A, accuracy ±0.5%;

[0084] Temperature sensor: NTC thermometer, measurement range -40~85℃, accuracy ±1℃;

[0085] BMS interface: Supports CAN2.0B protocol, sampling frequency 1kHz, and real-time acquisition of SOC data;

[0086] Power sensor: Employs a high-precision power transmitter with a measurement range of -1000~1000. Accuracy ±0.2%.

[0087] Data synchronization and preprocessing: Using a sampling period of 100ms (Ts), all sensor data are synchronized via timestamps to construct a multivariate design vector. ;in, Automatically determined by calculating the power change rate between adjacent sampling points: ;

[0088] in, The system automatically determines the power change rate by calculating the power change rate between adjacent sampling points.

[0089] 4. Model parameter optimization algorithm;

[0090] By comparing the shortcomings of traditional algorithms, an improved recursive least squares algorithm is proposed to achieve... Real-time optimization to address the issues of "poor real-time performance and data saturation" can include the following aspects:

[0091] 4.1. Scenario limitations of the traditional least squares method;

[0092] Traditional least squares methods employ a "full fit" approach, which suffers from a fatal technical bottleneck in real-time power compensation scenarios for energy storage power stations. The core issue lies in the inability to meet grid requirements in terms of real-time performance, specifically manifested as follows:

[0093] Full calculation results in a response lag; for each new power data point (energy storage sampling period is typically 100ms), the data needs to be reconstructed. Design matrix for each data ( (historical data volume) and observation vector and repeat the calculation →Matrix inversion →Matrix multiplication is a full operation.

[0094] For example, if a total of 6,000 data points are accumulated (approximately 10 minutes of sampling), the time taken for a single full-scale fitting is ≥500ms, which is much slower than the response window (100ms~300ms) for sudden changes in new energy output (such as a 20% drop in photovoltaic power within 1 second) or load fluctuations. This directly leads to a lag in the compensation power output, causing the grid voltage / frequency deviation to exceed the ±2% allowable range specified in the "Power Quality Supply Voltage Deviation" (GB / T12325-2008).

[0095] Core pain point: Inability to adapt to "millisecond-level compensation" requirements. The essence of traditional methods is that "adding one piece of data requires using all..." The computational load increases linearly with the amount of data, making it impossible to meet the real-time control requirement of "updating the compensation power once every 100ms" for energy storage power stations. This has become a key obstacle restricting energy storage from participating in the primary frequency regulation and inertia support of the power grid.

[0096] 4.2. Improved recursive least squares core logic;

[0097] The improvement of this invention for energy storage scenarios lies in the deep integration of "least square fitting" and "iterative update". Through the dual mechanism of "incremental update + forgetting factor decay", the model parameters (multivariate fitting coefficients) are optimized in real time. The core objective is to compress the time of a single parameter update to less than 20ms while ensuring fitting accuracy, adapt to the computing power of embedded controllers, and meet the millisecond-level compensation requirements.

[0098] Step 1: Gain Matrix Calculation - Balancing the Impact of New and Historical Data;

[0099] Gain matrix It is a quantitative indicator of "the magnitude of model correction by new data," which determines the extent to which each new sampled data can adjust the current model parameters. The calculation formula is: ;

[0100] in, : No. The covariance matrix at time step 1 records the "uncertainty in the current model parameter estimation". The larger the value, the lower the reliability of the current parameter, and more new data needs to be referenced.

[0101] : No. The multivariate design vector (6×1 dimension) at time step is in the form of It integrates new data features such as "basic power factor + battery status + fluctuation type";

[0102] Forgetting factor (typically 0.95~0.99), used to dynamically adjust the weight of historical data;

[0103] molecular : indicates the first The time-varying covariance matrix (reflecting the uncertainty of model parameters) and the first-order covariance matrix. The product of the multivariate design vectors at time step (reflecting the characteristics of the new data) is used to quantify the degree of correlation between the "uncertainty of the old model parameters" and the "characteristics of the new data". It is the core intermediate quantity for subsequent calculation of the gain matrix and determination of the magnitude of the correction of the model parameters by the new data.

[0104] denominator : Through the forgetting factor Achieving "weight normalization" The smaller the value, the faster the historical data weight decays, and the more sensitive the model is to new data (e.g., during shocks and fluctuations). ); The larger the value, the stronger the model stability (e.g., when taking a smooth working condition). ).

[0105] Breaking through the limitations of traditional least squares with "equal weighting of new and old data", the model dynamically allocates "correction intensity of new data" through the gain matrix, avoiding excessive interference from a single abnormal data (such as sensor false alarms) to the model, while ensuring a rapid response to effective new features (such as sudden changes in battery temperature).

[0106] Step 2: Time-varying update of the covariance matrix solves the problem of "data saturation";

[0107] covariance matrix It needs to be adjusted synchronously with parameter updates to reflect the "uncertainty of new parameters" in real time. At the same time, the "data saturation" problem of the traditional recursive least squares (RLS) method is solved by using a forgetting factor. The formula is as follows:

[0108] ;

[0109] Overcoming Data Saturation: Traditional RLS suffers from "continuous accumulation of historical data," leading to... As the data gradually approaches zero, the model loses its sensitivity to new data (i.e., "data saturation").

[0110] This invention is achieved through This item enables "long-term data to..." The "impact" decays exponentially—such as At that time, after 100 sampling periods, the old data pair The contribution is only the initial value This completely avoids the "fixation" of the model;

[0111] Dynamically adapting to time-varying operating conditions: When energy storage operating conditions change significantly (such as SOC-power curve drift caused by battery aging, or power characteristic degradation caused by low winter temperatures), It will automatically increase by updating the formula, improving the model's sensitivity to new data corrections and quickly adapting to new working conditions;

[0112] Numerical stability guarantee: The item is the "projection matrix", ensuring Always maintain symmetric positive definiteness to avoid the occurrence of singular matrices in numerical calculations and ensure the stability of parameter solutions.

[0113] Step 3: Iterative update of the coefficient vector to achieve a balance between timeliness and computational efficiency;

[0114] Based on the "prediction error of new data", and combined with the gain matrix, the current parameter vector is... (including) Instead of a full recalculation, incremental adjustments are made. The formula is as follows:

[0115] ;

[0116] in, : No. Measured power at any given time (real-time data collected from the energy storage grid connection point); Based on the first The "predictive power" of the time-matter model (the result of calculating the new data features using the old parameters);

[0117] "Prediction error" - The larger the difference, the more significant the deviation between the old model and the new operating conditions, and the more powerful the parameter correction is required.

[0118] Significantly improved computational efficiency: Each update requires only "vector multiplication + addition" operations, and the computational load is only related to the parameter dimension (6 dimensions). On an embedded MCU (such as STM32H7), the time taken is ≤20ms, which is far lower than the 500ms of traditional full computation, meeting the real-time requirement of a 100ms sampling period.

[0119] Prioritizing timeliness: The forgetting factor indirectly affects parameter updates through the gain matrix, making the "correction weight" of recent data much higher than that of long-term data (e.g., At that time, the correction contribution of the old data after 10 cycles was only 73.7% of the initial value, ensuring that the model always fits the current energy storage conditions.

[0120] 4.3. The specific steps are as follows:

[0121] Each time new data is collected (such as the first time...) Current per sampling period , After that, dynamically update based on the "deviation of the new data" (etc.). The formula consists of three steps, as shown below:

[0122] Step 1: First calculate the "gain matrix" "

[0123] ;

[0124] in: Previous cycle (number) The covariance matrix (given, for example, from) the time) or inherit);

[0125] : No. The periodic "multivariate design vector" (known, structure is) For example, after a certain collection );

[0126] Forgetting factor (known, ranging from 0.95 to 0.99; 0.97 is used in this invention);

[0127] : No. The periodic "multivariate design vector" (known, structure is) For example, after a certain collection );

[0128] Forgetting factor (known, ranging from 0.95 to 0.99; 0.97 is used in this invention);

[0129] : The transpose (converts a column vector into a row vector, e.g.) ).

[0130] Step 2: Calculate the updated covariance matrix. "

[0131]

[0132] in: The core function of the forgetting factor is to make the influence of historical data "decay exponentially" (e.g., hour, Each update weakens the weight of the old data.

[0133] It is still a 6×6 identity matrix (known);

[0134] Use the gain matrix to correct the "impact of new data on the old matrix" (the result is a 6×6 matrix).

[0135] Step 3: Finally, calculate the updated coefficient vector. "

[0136] formula ;

[0137] in, The coefficient vector of the previous cycle (kth time) is known and is a core parameter of the multivariate coupling model, with the following structure: For example, after the preorder iteration );

[0138] : The "measured power" for the (k+1)th cycle (known, collected from a high-precision power sensor at the energy storage grid connection point, unit: For example, after a certain collection );

[0139] The "predicted power" for the (k+1)th period (calculated using the old coefficient vector on the new data, unit: (This reflects the degree to which the old model fits the current operating conditions).

[0140] "Prediction error" (the difference between measured power and predicted power, unit: The larger the difference, the more significant the deviation between the old model and the current operating conditions, and the more powerful the coefficient correction is required.

[0141] A simplified example of an embodiment of the present invention is as follows (assuming a certain iteration):

[0142] Given:

[0143] Iteration After that, in the previous cycle ;

[0144] , ;

[0145] .

[0146] Step 1 calculation :

[0147] molecule = ;

[0148] Denominator = ;

[0149] so (6×1 vector).

[0150] Second step calculation :

[0151] ;

[0152] (A 6×6 matrix, with off-diagonal elements close to 0 and diagonal elements slightly less than 1).

[0153] Multiply by Divide by 0.97 to get the final result. — Diagonal elements will be larger than Slightly smaller (indicating that the coefficient estimates are more certain), while off-diagonal elements are still close to 0 (the correlation between coefficients is still weak).

[0154] Third step calculation Measured power Calculate the "predicted power":

[0155] ;

[0156] It should be noted that since the initial coefficients are empirical values, the prediction deviation in the first iteration is relatively large. In actual engineering, the initial K(0) can be optimized through the first three pre-iterations. In this example, the optimized predicted power is taken as 115kJ. To closely match the actual working conditions.

[0157] Calculate the "prediction error":

[0158] ;

[0159] Calculate the updated coefficient vector :

[0160] ;

[0161] Updated It can be directly substituted into the multivariate coupling model The target compensation power for the (k+1)th cycle is calculated to provide accurate parameters for subsequent SOC safety verification and PCS execution.

[0162] 5. Real-time power compensation control;

[0163] Closed-loop power control based on updated multivariable coefficients: Compensation power calculation: The compensation power at time k+1 is calculated by the following formula:

[0164] ;

[0165] The calculation results must be verified against security constraints to ensure that they are met:

[0166] ;

[0167] in The rated power of the energy storage converter is set to ensure system safety under extreme SOC conditions.

[0168] Power command output: The compensated power command, after safety verification, is sent to the energy storage converter, supporting continuous adjustment within the range of 0% to 100% of the rated power, with an adjustment step of ≤1%. The converter response time is ≤100ms, meeting the response requirements for primary frequency regulation in GB / T34120-2023.

[0169] In this embodiment of the invention, traditional univariate fitting models cannot reflect the coupled influence of multiple factors such as SOC, temperature, and power fluctuation type on the compensation power, resulting in low compensation accuracy, especially with errors exceeding ±5% under edge conditions. The fixed coefficient model of this application can adapt to time-varying characteristics such as battery aging and temperature drift, and its accuracy continues to improve after long-term operation. Conventional RLS algorithms suffer from data saturation problems and cannot effectively track rapid power fluctuations in the power grid. The solution of this application can meet the requirements of GB / T34120-2023 for primary frequency regulation response time (≤100ms) and regulation accuracy. It solves the problem of lacking a dynamic constraint mechanism for SOC and temperature and power output, which poses safety risks such as overcharging, over-discharging, and overheating power surges.

[0170] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that this embodiment is not limited to the described order of actions, because according to this embodiment, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to this embodiment.

[0171] Reference Figure 3 This diagram illustrates a structural block diagram of a real-time power compensation device for a multi-factor coupled energy storage power station according to this embodiment. Specifically, it may include the following modules:

[0172] Module 301 is used to build multivariate coupled models;

[0173] Initialization setting module 302 is used to perform parameter initialization settings for multivariable coupled models;

[0174] Acquisition module 303 is used for multi-dimensional data acquisition and preprocessing;

[0175] The update module 304 is used to optimize the multivariable coefficients of the multivariable coupled model in real time by calculating the gain matrix and iteratively updating the coefficient vector, so as to obtain the updated multivariable coefficients.

[0176] Control module 305 is used to implement closed-loop power control based on updated multivariable coefficients.

[0177] Preferably, the building module includes:

[0178] The acquisition submodule is used to acquire the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0179] A submodule is constructed to build a multivariate coupled model using the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0180] Preferably, the initialization setting module includes:

[0181] The first setting submodule is used to set the coefficient vector at the initial time as a multivariable coefficient vector;

[0182] The second settings submodule is used to set the covariance matrix initialization and forgetting factor.

[0183] Preferably, the acquisition module includes:

[0184] The acquisition submodule is used to achieve synchronous acquisition of multi-dimensional parameters using a distributed data acquisition architecture. It configures acquisition devices through current sensors, temperature sensors, BMS interfaces, and power sensors, with a sampling period of 100ms. It synchronizes all sensor data through timestamps to construct multivariate design vectors.

[0185] Preferably, the update module includes:

[0186] The determination submodule is used to determine the weights of dynamically adjusted historical data as the forgetting factor;

[0187] The calculation submodule is used to pass the first The covariance matrix at time t, the first The gain matrix is ​​calculated by using a multivariate design vector and forgetting factor at each time step.

[0188] Incremental correction submodule, used to pass the first The measured power at each time step, combined with the gain matrix, is used to incrementally correct the current parameter vector, and the coefficient vector is iteratively updated to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, resulting in the updated multivariable coefficients and covariance matrix. Adjustments are made synchronously as the multivariate coefficients are updated.

[0189] The modules in the aforementioned multi-factor coupled real-time power compensation device for energy storage power stations can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0190] The multi-factor coupled real-time power compensation device for energy storage power stations provided above can be used to execute the multi-factor coupled real-time power compensation method for energy storage power stations provided in any of the above embodiments, and has corresponding functions and beneficial effects.

[0191] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for a real-time power compensation device for a multi-factor coupled energy storage power station. The display screen can be a liquid crystal display (LCD) or an e-ink display. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0192] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0193] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above embodiment:

[0194] Construct a multivariate coupled model;

[0195] Perform parameter initialization settings for multivariable coupled models;

[0196] Multi-dimensional data collection and preprocessing;

[0197] By calculating the gain matrix and iteratively updating the coefficient vector, the multivariable coefficients of the multivariable coupled model are optimized in real time, and the updated multivariable coefficients are obtained.

[0198] Closed-loop power control is achieved based on the updated multivariable coefficients.

[0199] Preferably, the construction of the multivariate coupled model includes:

[0200] We obtained the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0201] A multivariate coupled model is constructed using the aforementioned basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0202] Preferably, the parameter initialization settings for the multivariable coupled model include:

[0203] Set the initial coefficient vector as a multivariate coefficient vector;

[0204] Set the covariance matrix initialization and forgetting factor.

[0205] Preferably, the multi-dimensional data acquisition and preprocessing includes:

[0206] A distributed data acquisition architecture is adopted to achieve synchronous acquisition of multi-dimensional parameters. The acquisition device is configured through current sensor, temperature sensor, BMS interface and power sensor. The sampling period is 100ms. All sensor data are synchronized by timestamp to construct multivariate design vector.

[0207] Preferably, the method of calculating the gain matrix and iteratively updating the coefficient vector to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, and obtaining the updated multivariable coefficients, includes:

[0208] The weights of dynamically adjusted historical data are determined as the forgetting factor;

[0209] Through the first The covariance matrix at time t, the first The gain matrix is ​​calculated by using a multivariate design vector and forgetting factor at each time step.

[0210] Through the first The measured power at each time step, combined with the gain matrix, is used to incrementally correct the current parameter vector, and the coefficient vector is iteratively updated to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, resulting in the updated multivariable coefficients and covariance matrix. Adjustments are made synchronously as the multivariate coefficients are updated.

[0211] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the above embodiments:

[0212] Construct a multivariate coupled model;

[0213] Perform parameter initialization settings for multivariable coupled models;

[0214] Multi-dimensional data collection and preprocessing;

[0215] By calculating the gain matrix and iteratively updating the coefficient vector, the multivariable coefficients of the multivariable coupled model are optimized in real time, and the updated multivariable coefficients are obtained.

[0216] Closed-loop power control is achieved based on the updated multivariable coefficients.

[0217] Preferably, the construction of the multivariate coupled model includes:

[0218] We obtained the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0219] A multivariate coupled model is constructed using the aforementioned basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

[0220] Preferably, the parameter initialization settings for the multivariable coupled model include:

[0221] Set the initial coefficient vector as a multivariate coefficient vector;

[0222] Set the covariance matrix initialization and forgetting factor.

[0223] Preferably, the multi-dimensional data acquisition and preprocessing includes:

[0224] A distributed data acquisition architecture is adopted to achieve synchronous acquisition of multi-dimensional parameters. The acquisition device is configured through current sensor, temperature sensor, BMS interface and power sensor. The sampling period is 100ms. All sensor data are synchronized by timestamp to construct multivariate design vector.

[0225] Preferably, the method of calculating the gain matrix and iteratively updating the coefficient vector to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, and obtaining the updated multivariable coefficients, includes:

[0226] The weights of dynamically adjusted historical data are determined as the forgetting factor;

[0227] Through the first The covariance matrix at time t, the first The gain matrix is ​​calculated by using a multivariate design vector and forgetting factor at each time step.

[0228] Through the first The measured power at each time step, combined with the gain matrix, is used to incrementally correct the current parameter vector, and the coefficient vector is iteratively updated to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, resulting in the updated multivariable coefficients and covariance matrix. Adjustments are made synchronously as the multivariate coefficients are updated.

[0229] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0230] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0231] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0232] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0233] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0234] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0235] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0236] The foregoing has provided a detailed description of a multi-factor coupled real-time power compensation method for an energy storage power station, a multi-factor coupled real-time power compensation device for an energy storage power station, a computer device, and a computer-readable storage medium provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A real-time power compensation method for a multi-factor coupled energy storage power station, characterized in that, The method includes: Construct a multivariate coupled model; Perform parameter initialization settings for multivariable coupled models; Multi-dimensional data collection and preprocessing; By calculating the gain matrix and iteratively updating the coefficient vector, the multivariable coefficients of the multivariable coupled model are optimized in real time, and the updated multivariable coefficients are obtained. Closed-loop power control is achieved based on the updated multivariable coefficients.

2. The real-time power compensation method for multi-factor coupled energy storage power stations according to claim 1, characterized in that, The construction of the multivariate coupled model includes: We obtained the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients. A multivariate coupled model is constructed using the aforementioned basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

3. The real-time power compensation method for multi-factor coupled energy storage power stations according to claim 1, characterized in that, The parameter initialization settings for the multivariable coupled model include: Set the initial coefficient vector as a multivariate coefficient vector; Set the covariance matrix initialization and forgetting factor.

4. The real-time power compensation method for multi-factor coupled energy storage power stations according to claim 1, characterized in that, The multi-dimensional data acquisition and preprocessing includes: A distributed data acquisition architecture is adopted to achieve synchronous acquisition of multi-dimensional parameters. The acquisition equipment is configured through current sensors, temperature sensors, BMS interfaces, and power sensors. With a sampling period of 100ms, all sensor data is synchronized through timestamps to construct a multivariate design vector.

5. The real-time power compensation method for a multi-factor coupled energy storage power station according to claim 1, characterized in that, The method of calculating the gain matrix and iteratively updating the coefficient vector enables real-time optimization of the multivariable coefficients of the multivariable coupled model, resulting in updated multivariable coefficients, including: The weights of dynamically adjusted historical data are determined as the forgetting factor; Through the first The covariance matrix at time t, the first The gain matrix is ​​calculated by using a multivariate design vector and forgetting factor at each time step. Through the first The measured power at each time step, combined with the gain matrix, is used to incrementally correct the current parameter vector, and the coefficient vector is iteratively updated to achieve real-time optimization of the multivariable coefficients of the multivariable coupled model, resulting in the updated multivariable coefficients and covariance matrix. Adjustments are made synchronously as the multivariate coefficients are updated.

6. A real-time power compensation device for a multi-factor coupled energy storage power station, characterized in that, The device includes: Modules for building multivariate coupled models; The initialization settings module is used to initialize parameters for multivariable coupled models. The data acquisition module is used for multi-dimensional data acquisition and preprocessing. The update module is used to optimize the multivariate coefficients of a multivariate coupled model in real time by calculating the gain matrix and iteratively updating the coefficient vector, so as to obtain the updated multivariate coefficients. The control module is used to implement closed-loop power control based on the updated multivariable coefficients.

7. The real-time power compensation device for a multi-factor coupled energy storage power station according to claim 6, characterized in that, The building module includes: The acquisition submodule is used to acquire the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients. A submodule is constructed to build a multivariate coupled model using the basic power influence factor, battery state of charge, battery temperature, power fluctuation type identifier, and multivariate fitting coefficients.

8. The real-time power compensation device for a multi-factor coupled energy storage power station according to claim 6, characterized in that, The initialization setting module includes: The first setting submodule is used to set the coefficient vector at the initial time as a multivariable coefficient vector; The second settings submodule is used to set the covariance matrix initialization and forgetting factor.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the real-time power compensation method for multi-factor coupling energy storage power stations as described in any one of claims 1 to 5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the real-time power compensation method for multi-factor coupling energy storage power stations as described in any one of claims 1 to 5.

Images