A method for seamless grid-connected and off-grid switching of grid-connected energy storage UPS

Abstract

This invention discloses a seamless grid-connected / off-grid switching method for a grid-connected energy storage UPS, belonging to the field of grid-connected / off-grid switching technology. The method includes: real-time identification of the startup time point; performing degradation analysis based on the startup time point to obtain a verification period; performing a seamless grid-connected / off-grid switching self-test based on the verification period, and real-time monitoring of transient voltage, frequency, phase, and load power supply status during switching to evaluate the switching effect from multiple dimensions and obtain the self-test results; performing optimization processing based on the self-test results and generating an optimized control strategy; writing the optimized control strategy into the UPS grid-connected control kernel in real time to achieve adaptive iterative upgrades of the control strategy; performing seamless grid-connected / off-grid switching control; abandoning the traditional UPS manual, dedicated debugging, and simple self-test mode that only verifies battery switching, the method uses an autonomous, seamless self-verification mechanism under normal load operation to automatically select the optimal time period to complete the full performance verification of grid-connected / off-grid switching without manual intervention.

Description

Technical Field

[0001] This invention belongs to the field of grid-connected and off-grid switching technology, specifically a method for seamless grid-connected and off-grid switching of a grid-type energy storage UPS. Background Technology

[0002] Network-based energy storage UPS systems, with their active voltage build-up and frequency regulation capabilities, are widely used in critical power supply scenarios such as data centers and industrial precision loads. Their seamless switching capability between grid-connected and off-grid modes is a core indicator for ensuring uninterrupted power supply and mitigating electrical disturbances. While current network-based energy storage UPS products on the market possess basic self-testing functions, these modes only support manual operation during the commissioning phase or simple battery button self-testing. A typical operation involves pressing and holding the function button to force a switch to battery power and then switching back to mains power, only providing basic verification of battery energy storage switching and inverter startup functions. Furthermore, existing products completely lack an autonomous, periodic, and seamless switching self-verification mechanism under normal load operation. This prevents automatic switching performance verification during unattended periods (such as off-peak hours at night), leading to potential switching performance degradation and the inability to detect latent switching faults over long-term operation. This can result in switching failures and power supply fluctuations when real grid faults trigger the switching process.

[0003] Based on this, in order to achieve efficient, accurate, and safe seamless switching between grid-connected and off-grid, this invention provides a method for seamless switching between grid-connected and off-grid of a grid-type energy storage UPS. Summary of the Invention

[0004] To address the problems of the above solutions, this invention provides a method for seamless switching between grid-connected and off-grid operation of a grid-connected energy storage UPS.

[0005] The objective of this invention can be achieved through the following technical solutions: A method for seamless grid-connected and off-grid switching of a grid-connected energy storage UPS includes the following steps: S1: Real-time identification of startup time point, and degradation analysis based on startup time point to obtain the verification period; Furthermore, determining the start time includes: The system presets a verification period and identifies the verification time, optimization time, and grid connection / offline switchover time closest to the current time. It then selects the verification time, optimization time, or grid connection / offline switchover time closest to the current time as the base time. The start time is determined by adding the verification period to the base time.

[0006] Furthermore, determining the start time includes: Determining the start time includes: The system presets a verification period and identifies the verification time, optimization time, and grid connection / offline switchover time closest to the current time. It then selects the verification time, optimization time, or grid connection / offline switchover time closest to the current time as the base time. The first time point is calculated based on the verification cycle and the base time. Obtain predetermined standard parameters for the design, including preset allowable voltage fluctuation values, frequency offset thresholds, phase synchronization error upper limits, and rated switching transition times, to form a standard design sample set; simultaneously, establish a dedicated self-test timing database to store measured performance parameters for self-testing; and form a continuous timing sample sequence. The predicted switching performance parameters for the corresponding prediction time are obtained by analyzing the preset performance degradation prediction model. The overall degradation degree is calculated based on the degradation degree prediction formula. Degradation prediction formula: ; In the formula: To determine the overall degradation level, n represents the total number of performance evaluation metrics, which is 4; k=1: voltage fluctuation; k=2: frequency offset; k=3: phase synchronization error; k=4: switching response time; To predict switching performance parameters; Y 0k Predetermined standard parameters for design; The time point when the overall degradation degree is greater than the threshold X1 is marked as the second time point; Compare the first time point with the second time point, and select the time point closest to the current time as the start time point.

[0007] Furthermore, the third time point is calculated, and the first, second, and third time points are compared. The time point closest to the current time is selected as the start time point. The calculation of the third time point includes: Acquire target-related data; calibrate the overall degradation degree at the corresponding time using the target-related data; When the corresponding comprehensive degradation levels meet the on-grid and off-grid handover performance requirements of the target associated data, the third time period is determined to be in a pending state. When the overall degradation level does not meet the on-grid / off-grid handover performance requirements of the target associated data, a third time point is determined based on the corresponding time period of the target associated data and the preset preparation time period.

[0008] Furthermore, the start-up time and verification period are prohibited during peak load periods and periods of power grid anomalies.

[0009] Furthermore, degradation analysis is performed based on the startup time, including: Real-time acquisition of multi-dimensional operating parameters, including operating parameters of grid-type energy storage UPS, power grid operating parameters and load operating parameters; By analyzing multi-dimensional operating parameters using a pre-set multi-factor analysis model, the optimal silent verification period for dynamic equipment under load operation is obtained.

[0010] Furthermore, the expression for the multifactor analysis model is: ; In the formula: S(t) represents the fitness of the corresponding time. ω represents the standardized and normalized range of each core factor, taking values ​​in the range [0, 1]. i (t) represents the dynamic adaptive weights calculated based on real-time operating condition information entropy, which are updated in real-time throughout the process and satisfy... ; α is the nonlinear coupling term of the core factors of grid voltage and frequency, which characterizes the grid synchronization correlation constraint; α is the coupling term adjustment coefficient, which is used to control the contribution ratio of the coupling term to the overall score; β is the operating condition over-limit penalty factor; η(t) is the time series sliding integral stability coefficient, which is calculated by the ratio of the deviation between the mean integral value of the time series data in the past 10 minutes and the instantaneous value, and filters out the misjudgment interference caused by instantaneous grid jitter and short-term load changes. Set verification time period standards, and determine the verification time period based on the verification time period standards and the adaptability of each time period.

[0011] Furthermore, the value of α ranges from [0.1, 0.5].

[0012] S2: Perform seamless and off-grid switching self-tests according to the verification period, and monitor the switching transient voltage, frequency, phase and load power supply status in real time to evaluate the switching effect from multiple dimensions and obtain the self-test results.

[0013] S3: Optimize based on self-test results and generate optimized control strategies.

[0014] Furthermore, an optimized time is generated after optimization processing.

[0015] S4: The optimized control strategy is written into the UPS network control kernel in real time, replacing the original baseline control parameters, and realizing the adaptive iterative upgrade of the control strategy.

[0016] S5: Performs seamless switching control between grid connection and off-grid operation.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention abandons the traditional UPS's simple self-test mode, which requires manual operation, dedicated debugging, and verification of only battery switching. Through an autonomous and imperceptible self-verification mechanism under normal load operation, it can automatically select the optimal time period to complete and perform full-performance verification of off-grid switching without manual intervention or affecting normal load operation. It can detect hidden switching faults and performance degradation problems that traditional self-tests cannot identify in advance, significantly improving the reliability of equipment operation. Combined with simulation-based pre-judgment dual protection, it avoids invalid self-tests and switching disturbances during peak load periods and abnormal grid periods, solving the drawbacks of traditional manual self-tests, which are arbitrary in timing and have uncontrollable risks. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0019] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] The technical solution of the present invention will be clearly and completely described below with reference to 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, a method for seamless grid-connected and off-grid switching of a grid-connected energy storage UPS includes the following steps: S1: Real-time identification of the start time point, where the start time point indicates the verification period determined after the corresponding time, i.e., the subsequent degradation analysis begins from this point, and is used to prompt for degradation analysis; degradation analysis is performed based on the start time point to obtain the verification period; In one embodiment, the start time point can be identified by a periodic time period, such as calculating the start time point based on the previous verification time or optimization time and a preset period. The optimization time refers to the time for optimization and adjustment after subsequent evaluations fail.

[0022] For example, a preset verification period, such as 24 hours, 3 days, 7 days, etc., is used to identify the previous verification time and optimization time, select the previous verification time or optimization time that is closest to the current time as the base time, and add the verification period to the base time to calculate the start time.

[0023] In one embodiment, if there is a grid connection / off-grid switchover in the middle, the time of that switchover is used as the base time, that is, the base time closest to the current time is selected from the previous verification time, optimization time, and grid connection / off-grid switchover time.

[0024] In one embodiment, determining the start time point includes: The system presets a verification period and identifies the verification time, optimization time, and grid connection / offline switchover time closest to the current time. It then selects the verification time, optimization time, or grid connection / offline switchover time closest to the current time as the base time. The first time point is calculated based on the verification cycle and the base time. Obtain the seamless switching baseline indicators (design-defined standard parameters) predetermined by the factory design, including preset voltage fluctuation allowable values, frequency offset thresholds, phase synchronization error upper limits, and switching transition time ratings, to form a standard design sample set; simultaneously, establish a dedicated self-test timing database, continuously storing the measured performance parameters of recent (e.g., the last 30 self-tests), forming a continuous timing sample sequence, and updating it again based on this as a benchmark for cases where optimization processing has been performed; Y k (1), Y k (2), ..., Y k (N), where N is the number of historical self-checks, Y k These are the switching performance metrics corresponding to each self-test. A performance degradation prediction model is constructed using a weighted time-series multinomial fitting algorithm. Recent validation data are used to fit the performance change trend and predict future performance parameter switching times. The prediction formula is as follows: ; In the formula: The predicted switching performance parameters are: τ0, τ1, and τ2 are time series fitting coefficients, obtained iteratively from historical self-test data; λ is the trend correction coefficient; σ i The weights are assigned based on time sequence, following the principle of "nearer data has higher weights, farther data has lower weights," thus reinforcing the impact of the latest performance changes; Y k (i) represents the measured parameters of the i-th historical self-test, Y 0k Pre-defined standard parameters for design.

[0025] Based on the predicted switching performance parameters, the overall degradation degree for future periods is further calculated to complete the trend prediction. The degradation degree prediction formula is as follows: ; In the formula: To represent the overall degradation level, n is the total number of performance evaluation index dimensions, which is 4; the number of the four core performance evaluation indicators, k=1: voltage fluctuation; k=2: frequency offset; k=3: phase synchronization error; k=4: switching response time; Mark the time point when the overall degradation degree is greater than the threshold X1 as the second time point; if the threshold X1 is 8%, it is generally set to 8%~15%, depending on the actual needs; Compare the first time point with the second time point, and select the time point closest to the current time as the start time point.

[0026] In one embodiment, the performance degradation prediction model or the determination of the second time point can be established based on machine learning, deep learning algorithms, etc. For example, a training set can be set manually. The training set includes input data and output data. The input data is the seamless switching benchmark index and multiple recent sets of validation data. The output data is the predicted switching performance parameters for the corresponding prediction time. The training is performed using the training set.

[0027] In one embodiment, the difference between this embodiment and the previous embodiment is that a third time point is determined, the first time point, the second time point and the third time point are compared, and the time point closest to the current time is selected as the start time point.

[0028] The determination of the third time point includes: Acquire target-related data that affects the performance requirements of grid-connected / off-grid switching, such as production content and management needs; for example, production scheduling (to understand the performance requirements of products produced within the corresponding time period for grid-connected / off-grid switching); calibrate the overall degradation level of the corresponding time period using the target-related data to determine whether the overall degradation level of the corresponding time period meets the requirements of the grid-connected / off-grid switching performance of the target-related data. When the calibration is successful (the requirements are met), the third time is in a pending state. It can also be represented in other ways, such as A, which means greater than the first time point and the second time point, or it can be represented as infinity, etc. When calibration fails, a third time point is determined based on the corresponding time of the relevant target-related data. The third time point is the time of the preset preparation period, which is based on the start of the time period corresponding to the target-related data. This is because a period of time needs to be reserved to determine the verification period, so as to complete the verification before the time period corresponding to the target-related data and to optimize in advance if the target fails to meet the standard. For example, if product A is processed from August 1st to August 10th, according to production requirements and customer needs, it is necessary to ensure that the performance of the off-grid switching reaches the design performance during this period, with an allowable deviation of 2%. Then, if the overall degradation is less than or equal to 2%, the calibration passes; otherwise, the calibration fails. If it fails, the third time point is the preset time period starting from August 1st. If the preset preparation period is 1 week, then the third time point is 0:00 on July 24th.

[0029] In one embodiment, mandatory requirements may also be added, such as prohibiting peak load periods and setting verification periods during abnormal power grid periods.

[0030] In one embodiment, a minimum self-test interval can be preset, meaning that the time interval between adjacent self-tests cannot be less than the minimum self-test interval.

[0031] In one embodiment, degradation analysis based on the startup time point includes: Real-time acquisition of operating parameters of grid-type energy storage UPS, grid operating parameters and load operating parameters, completion of data preprocessing and feature extraction, providing data support for self-verification period determination and switching performance degradation prediction, and integration into multi-dimensional operating parameters; By using a pre-set multi-factor analysis model to analyze multi-dimensional operating parameters, the optimal silent verification period for equipment under load operation is dynamically selected.

[0032] In one embodiment, the multi-factor analysis model can be built based on machine learning and deep learning algorithms, and trained by manually setting a training set. The training set includes input data and output data, where the input data are multi-dimensional operating parameters and the output data is the validation period.

[0033] In one embodiment, the establishment of a multi-factor analysis model is as follows: Set core factors: grid voltage fluctuation ΔU, grid frequency offset Δf, battery SOC margin, load power fluctuation PL, equipment operating temperature T, and historical switching failure coefficient F. Each core factor is normalized by range standardization to eliminate dimensional differences, mapping each factor to the [0, 1] interval to obtain the core parameters of each core factor. The normalization formula is as follows: ; In the formula: x i (t) The real-time sampled value of the i-th core factor, x i，max and x i，min The maximum and minimum values ​​of the i-th core factor under historical operating conditions or within a preset safety range; Based on this, a dynamic adaptive weight ω is introduced, which is calculated based on the entropy of real-time operating conditions. i (t), the dynamically adaptive weights are updated in real time throughout the process and satisfy ∑ω i (t)=1; The uncertainty of each core factor is quantified by the entropy value of the operating condition data. The operating condition factor with greater fluctuation has a higher real-time weight, thereby achieving priority constraint on key operating conditions. The calculation process of dynamic entropy weight is as follows: First, based on a sliding window of length T, the probability distribution of the i-th core factor within the window is calculated: k=0,1,…,T-1; k is the temporal offset index within the sliding window, representing the sampling time sequence number within the window. k=0: the data of the latest time t within the window; k=1: the historical time t−1 pushed back by 1 step; which is different from k in the above embodiment.

[0034] Next, calculate the information entropy of the i-th core factor: ; Then, the difference coefficient of the i-th core factor is calculated: ; Finally, normalization yields the dynamic weights: ; Based on this, a multi-factor analysis model is constructed by introducing factor cross-coupling coefficient, operating condition over-limit penalty factor, time series sliding integral coefficient, and information entropy dynamic weight. The overall calculation formula is as follows: ; In the formula: S(t) represents the adaptability at the corresponding time, which is equivalent to a comprehensive quantitative score of whether the corresponding time is suitable for performing a seamless and offline handover self-test. ω represents the standardized and normalized range of each core factor, taking values ​​in the range [0, 1]. i (t) represents the dynamic adaptive weights calculated based on real-time operating condition information entropy, which are updated in real-time throughout the process and satisfy... ; The term represents the nonlinear coupling of the core factors of grid voltage and frequency. It characterizes the grid synchronization correlation constraints, simulates the strong voltage-frequency coupling electrical characteristics during real grid-connected and off-grid switching, and fills the gap in the linear model lacking physical coupling. α is the coupling term adjustment coefficient, ranging from [0.1, 0.5], used to control the contribution of the coupling term to the overall score. β is the operating condition over-limit penalty factor, employing a piecewise hard constraint mechanism. When any core parameter exceeds the safety threshold, it automatically and significantly lowers the overall score, preventing self-checks during edge risk periods. Specifically, it takes the following form: ; The recommended values ​​are β1 = 0.3~0.5, β2 = 0 or a minimum value close to 0; η(t) is the time-series sliding integral stability coefficient, calculated by the ratio of the deviation between the mean of the time-series data over the past 10 minutes and the instantaneous value, filtering out misjudgment interference caused by instantaneous grid jitter and short-term load changes. The formula for calculating the mean of the time-series sliding integral is: ; The formula for calculating the time-series sliding integral stability coefficient is: ; Where γ is the deviation adjustment coefficient, with a value range of [1, 2], controlling the degree of influence of deviation on the coefficient. When the deviation ratio exceeds the preset threshold (e.g., 20%), η(t) is forced to = 0, and it is directly judged as a non-steady-state period. At the same time, this model adopts a hierarchical dynamic threshold + continuous steady-state judgment logic. Under normal stable working conditions, the judgment threshold S0 = 0.85, and under weak fluctuation complex working conditions, it is adaptively lowered to S0 = 0.80. It must also meet the requirement that the time series integral adaptation is stable and meets the standard without positive or negative jumps for 10 consecutive sampling periods (the rolling calculation period of the algorithm score) before it can be locked as the optimal imperceptible verification period (verification period standard). After the self-check is completed, the system updates the entropy weight matrix and coupling coefficient in reverse based on the current working condition data to realize the self-learning iteration of model parameters.

[0035] S2: Perform seamless and off-grid switching self-tests according to the verification period, and monitor the switching transient voltage, frequency, phase and load power supply status in real time to evaluate the switching effect from multiple dimensions and obtain the self-test results.

[0036] In one embodiment, a multi-dimensional switching effect evaluation can be performed by calculating the overall degradation degree according to the above embodiment, evaluating the effect based on the overall degradation degree, and determining whether it is qualified and whether optimization processing is required.

[0037] In one embodiment, a multi-dimensional switching effect evaluation can be performed, or the effect evaluation can be based on other existing methods. This can identify hidden switching deviations and performance degradation that traditional self-inspection cannot detect, automatically adjust model weight parameters and switching closed-loop control coefficients to address evaluation defects, continuously optimize switching performance, and achieve normalized autonomous verification and dynamic adaptation optimization.

[0038] S3: Optimize based on self-test results and generate optimization time and optimization control strategy.

[0039] Most performance degradation during switching does not require system shutdown or disassembly for maintenance. Only severe hardware aging and degradation scenarios trigger maintenance prompts, in which case the relevant personnel should be notified to handle the issue.

[0040] For other situations, an optimized control strategy is determined, and optimization is completed online with full load, including fine-tuning of grid-connected synchronous phase-locked loop parameters, correction of grid-connected phase compensation coefficient, tuning of off-grid voltage amplitude and frequency regulation parameters, adaptive calibration of switching dead zone threshold, iteration of PID dynamic adjustment parameters, updating of entropy weight matrix and coupling coefficient of multi-factor analysis model, correction of time series prediction fitting coefficient, and optimization of parameters grouped in operating condition intervals.

[0041] Based on historical data, the core causes of performance degradation when self-inspection results are unqualified can be identified, and corresponding optimization control strategies can be determined. For example, the K-means clustering algorithm can be used to divide the power grid into multiple operating conditions such as steady-state power grid, fluctuating power grid, light load, and heavy load, and to trace and locate the core causes of performance degradation.

[0042] For example, for mild degradation, phase-locked loop phase compensation fine-tuning is performed; for moderate degradation, dynamic correction of grid voltage amplitude and frequency modulation coefficient is performed; for severe degradation, switching logic parameters are reset and high-risk switching conditions are locked. For example, the overall degradation rate in the range (3%, 8%) is slow mild degradation, in the range (8%, 15%) is continuous moderate degradation, and greater than 15% is rapid severe degradation.

[0043] S4: The optimized control strategy is written into the UPS network control kernel in real time, replacing the original baseline control parameters, and realizing the adaptive iterative upgrade of the control strategy.

[0044] S5: Performs seamless switching control between grid connection and off-grid operation.

[0045] For example, when the system detects a power grid fault, an abnormal mains power supply, or receives an active off-grid instruction from the background EMS, it initiates seamless off-grid handover; During off-grid operation, the system is completely disconnected from the mains power grid. The grid-type energy storage UPS independently constructs a stable AC voltage and frequency reference to continuously ensure uninterrupted power supply to the load.

[0046] When the mains power grid parameters are detected to have returned to normal and remain stable, or when a proactive grid connection instruction is received from the background, the pre-synchronous grid connection switch is initiated, and the grid connection reset is completed without any impact or interruption.

[0047] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a grid-type energy storage UPS seamless switching method between grid and off-grid as described in the above embodiments.

[0048] The above formulas are all numerical calculations after removing dimensions. The formulas are obtained by software simulation based on a large amount of data and are closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.

[0049] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for seamless switching between grid-connected and off-grid operation of a grid-connected energy storage UPS, characterized in that, Includes the following steps: S1: Real-time identification of startup time point, and degradation analysis based on startup time point to obtain the verification period; Determining the start time includes: The system presets a verification period and identifies the verification time, optimization time, and grid connection / offline switchover time closest to the current time. It then selects the verification time, optimization time, or grid connection / offline switchover time closest to the current time as the base time. The first time point is calculated based on the verification cycle and the base time. Obtain predetermined standard parameters for the design, including preset allowable voltage fluctuation values, frequency offset thresholds, phase synchronization error upper limits, and rated switching transition times, to form a standard design sample set; at the same time, establish a dedicated self-test timing database to store the measured performance parameters of the self-test; and form a continuous timing sample sequence. The predicted switching performance parameters for the corresponding prediction time are obtained by analyzing the preset performance degradation prediction model. The overall degradation degree is calculated based on the degradation degree prediction formula. The degradation degree prediction formula is as follows: ； In the formula: To determine the overall degradation level, n represents the total number of performance evaluation metrics, which is 4; k=1: voltage fluctuation; k=2: frequency offset; k=3: phase synchronization error; k=4: switching response time; To predict switching performance parameters; Y 0k Predetermined standard parameters for design; The time point when the overall degradation degree is greater than the threshold X1 is marked as the second time point; Compare the first time point with the second time point, and select the time point closest to the current time as the start time point; Degradation analysis is performed based on the startup time to obtain the verification period, including: Real-time acquisition of multi-dimensional operating parameters, including operating parameters of grid-type energy storage UPS, power grid operating parameters and load operating parameters; By analyzing multi-dimensional operating parameters through a preset multi-factor analysis model, the optimal silent verification period under the dynamic equipment load operation state is obtained, which is used as the verification period. The expression for the multifactor analysis model is: ； In the formula: S(t) represents the fitness of the corresponding time, and t is time; ω represents the standardized and normalized range of each core factor, taking values ​​in the range [0, 1]. i (t) represents the dynamic adaptive weights calculated based on real-time operating condition information entropy, which are updated in real-time throughout the process and satisfy... ; α is the nonlinear coupling term of the core factors of grid voltage and frequency, which characterizes the grid synchronization correlation constraint; α is the coupling term adjustment coefficient, with a value range of [0.1, 0.5], used to control the contribution ratio of the coupling term to the overall score; β is the operating condition over-limit penalty factor, and η(t) is the time-series sliding integral stability coefficient; Set verification time period standards, and determine the verification time period based on the verification time period standards and the adaptability of each time period; S2: Perform seamless and off-grid handover self-test according to the verification period, and monitor the handover transient voltage, frequency, phase and load power supply status in real time to evaluate the handover effect from multiple dimensions and obtain the self-test results; S3: Optimize based on self-test results and generate optimized control strategies; S4: The optimized control strategy is written into the UPS network control kernel in real time, replacing the original baseline control parameters, and realizing the adaptive iterative upgrade of the control strategy; S5: Performs seamless switching control between grid connection and off-grid operation.

2. The method for seamless grid-connected / off-grid switching of a grid-type energy storage UPS according to claim 1, characterized in that, Another way to determine the start time includes: The system presets a verification period and identifies the verification time, optimization time, and grid connection / offline switchover time closest to the current time. It then selects the verification time, optimization time, or grid connection / offline switchover time closest to the current time as the base time. The start time is determined by adding the verification period to the base time.

3. The method for seamless grid-connected / off-grid switching of a grid-type energy storage UPS according to claim 1, characterized in that, Calculate the third time point, compare the first, second, and third time points, and select the one closest to the current time as the start time point; The calculation of the third time point includes: Acquire target-related data; calibrate the overall degradation degree at the corresponding time using the target-related data; When the corresponding comprehensive degradation levels meet the on-grid and off-grid handover performance requirements of the target associated data, the third time period is determined to be in a pending state. When the overall degradation level does not meet the on-grid / off-grid handover performance requirements of the target associated data, a third time point is determined based on the corresponding time period of the target associated data and the preset preparation time period.

4. The method for seamless grid-connected / off-grid switching of a grid-type energy storage UPS according to claim 1, characterized in that, The start-up time and verification period are prohibited during peak load periods and periods of power grid abnormality.

5. The method for seamless grid-connected / off-grid switching of a grid-type energy storage UPS according to claim 1, characterized in that, The optimized time is generated after optimization processing.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of a grid-type energy storage UPS seamless switching method for grid-connected and off-grid operation as described in any one of claims 1 to 5.

Images