A control system for off-grid parallel connection of energy storage inverters

Abstract

The application relates to the technical field of power grid control, and discloses a control system for off-grid parallel connection of an energy storage inverter, which comprises the following modules: an off-grid parallel connection judgment module for judging whether to enter an off-grid parallel connection mode; an off-grid time point determination module for determining an off-grid time point of a power grid; a power consumption calculation module for calculating real power consumption; a battery charging optimization module for performing charging optimization on a slave battery; and a battery off-grid parallel connection module for synchronously supplying power to a load end after available batteries are off-grid and connected in parallel; the application can accurately capture a critical time point at which a real-time state of the power grid changes significantly and irreversibly, eliminates interference factors caused by normal fluctuations of the power grid, avoids problems of mis-triggering or delayed response of off-grid parallel connection operation caused by a single threshold value or simple fluctuation judgment, thereby providing a reliable time reference for subsequent calculation of an off-grid time length and estimation of real power consumption, guaranteeing accuracy of off-grid parallel connection control in time sequence logic, and improving reliability of off-grid parallel connection operation.

Description

Technical Field

[0001] This invention relates to the field of power grid control technology, and more specifically, to a control system for off-grid parallel connection of energy storage inverters. Background Technology

[0002] As a key power electronic device connecting energy storage batteries with the grid and loads, energy storage inverters have been widely used in microgrids, residential energy storage, and industrial and commercial energy storage scenarios. Energy storage inverters typically have two operating modes: grid-connected power supply mode and off-grid power supply mode. When the grid experiences operational failures, voltage drops, or frequency anomalies, the energy storage inverter needs to switch from grid-connected mode to off-grid mode to ensure uninterrupted power supply to the load.

[0003] Reference patent application CN114362229A discloses a control system for off-grid parallel operation of energy storage inverters, including a power supply control unit and a load monitoring unit. The power supply control unit is configured to monitor the energy storage status of the energy storage units and control the power output of the energy storage units. The load monitoring unit includes monitoring sub-units coupled to multiple loads and a central control unit communicatively connected to the multiple monitoring sub-units. The monitoring sub-units collect the operating parameters of the electrical loads and establish a time-series-based power consumption model to predict the future power demand of the loads and formulate power consumption strategies. By monitoring the energy storage status of the energy storage units and combining it with the power consumption strategies, the power supply strategy is determined, thereby controlling the parallel operation of at least two energy storage units after off-grid operation. Existing off-grid parallel control systems determine the precise timing of off-grid parallel operation by identifying whether the fluctuation amplitude of characteristic parameters in the power grid exceeds the controllable range. However, they lack in-depth analysis methods for the geometric characteristics of the changing trends of characteristic parameters and the ability to dynamically track the process of power grid state changes. As a result, it is difficult to accurately capture the precise critical moment when the power grid transitions from an online state to an off-grid state. Consequently, the timing of the off-grid switching of the energy storage inverter is not precise and reliable enough, reducing the timeliness of subsequent power supply strategies.

[0004] In view of this, the present invention proposes a control system for off-grid parallel connection of energy storage inverters to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a control system for off-grid parallel operation of an energy storage inverter, applied to an off-grid parallel platform, comprising: The off-grid parallel connection determination module is used to collect the online power supply characteristics of the power grid during the control period, predict the power grid status in the next control period through the power grid status identification model, and determine whether the energy storage inverter enters the off-grid parallel connection mode based on the power grid status. The off-grid time determination module is used to perform time-granularity analysis and dynamic sliding window tracking of online power supply characteristics, draw trend curves, extract geometric features of trend curves, and determine the off-grid time of the power grid based on the geometric features. The power consumption calculation module is used to collect the load power consumption parameters of the load end during the control period, predict the theoretical power consumption of the load end in the next control period through the power consumption prediction model, determine the off-grid duration based on the off-grid time, and calculate the actual power consumption by combining the off-grid duration and the theoretical power consumption. The battery charging optimization module is used to obtain the remaining battery capacity and battery voltage of the energy storage battery, divide the energy storage battery into master battery and slave battery according to the remaining battery capacity, and optimize the charging of master battery and slave battery according to the battery voltage. The off-grid parallel battery module is used to select available batteries from energy storage batteries based on a dual parallel mechanism, and then connect the available batteries off-grid in parallel to synchronously supply power to the load.

[0006] Furthermore, online power supply characteristics include voltage sequence, frequency sequence, power balance sequence, and device status sequence; The power grid status includes online status and offline status. The online status is converted to 11, and the offline status is converted to 00. When the output of the power grid status identification model is 11, the power grid status is online; when the output of the power grid status identification model is 00, the power grid status is offline.

[0007] Furthermore, when the grid is in an online state, the energy storage inverter is determined not to enter the off-grid parallel mode; when the grid is in an offline state, the energy storage inverter is determined to enter the off-grid parallel mode.

[0008] Furthermore, the trend curves include voltage curves, frequency curves, power curves, and device curves; The method for drawing the trend curve is as follows: Using the displayed values ​​of voltage, frequency, power balance, and equipment status at all times as the vertical axis and time as the horizontal axis, coordinate points are marked in a two-dimensional coordinate system to generate voltage coordinate system, frequency coordinate system, power coordinate system, and equipment coordinate system. In the voltage coordinate system, frequency coordinate system, power coordinate system, and equipment coordinate system, two coordinate points at adjacent positions are connected sequentially by a smooth curve to draw the voltage curve, frequency curve, power curve, and equipment curve.

[0009] Furthermore, geometric features include slope abrupt change points and curvature abrupt change points; The method for extracting geometric features is as follows: In the voltage curve, frequency curve, power curve, and equipment curve, the slope of the tangent is calculated by subtracting the displayed value of the next coordinate point from the displayed value of the previous coordinate point and dividing by the displayed value of the previous coordinate point. According to the chronological order, all tangent slopes are summarized into a slope sequence, and the slope sequence is divided into continuously distributed slope windows based on the preset window duration. In the slope window, the slope difference is calculated by subtracting the slope of the last tangent from the slope of the first tangent, and the coordinates of the points where the slope difference is greater than the preset slope difference threshold are recorded as slope change points. Starting from the second coordinate point, the three coordinate points in adjacent positions are denoted as a point set, and the curvature of the curve of the point set is calculated using the curvature formula; According to the chronological order, all curve curvatures are summarized into a curvature sequence, and the curvature sequence is divided into continuously distributed curvature windows based on the preset window duration. In the curvature window, the curvature difference is calculated by subtracting the curvature of the last curve from the curvature of the first curve. The set of points whose curvature difference is greater than the preset curvature difference threshold is recorded as the mutation set, and the coordinate point in the middle of the mutation set is recorded as the curvature mutation point.

[0010] Furthermore, the method for determining the off-network time is as follows: In the voltage curve, frequency curve, power curve, and equipment curve, the rated voltage line and rated frequency line are plotted respectively, using the rated voltage and rated frequency as the standard. Along the vertical axis, calculate the first vertical distance between the slope abrupt change point and the rated voltage line, and between the slope abrupt change point and the rated frequency line, and record the slope abrupt change point corresponding to the minimum value of the first vertical distance as a type of abrupt change point. Along the vertical axis, calculate the second longitudinal distance between the curvature abrupt change point and the rated voltage line, and between the curvature abrupt change point and the rated frequency line, and record the curvature abrupt change point corresponding to the minimum value of the second longitudinal distance as the second type of abrupt change point; Arrange all the Class I and Class II mutation points into a mutation queue according to the chronological order. The time corresponding to the last Class I or Class II mutation point in the mutation queue is recorded as the off-network time.

[0011] Furthermore, the load's electrical parameters include temperature, humidity, light intensity, time of day, and load power; The actual electricity consumption is calculated as follows: The current time is taken as the start time of the next control period, and the duration between the start time and the offline time is recorded as the online duration; The offline time is calculated by subtracting the duration of the control period from the online time. The offline time is then compared with the duration of the control period and combined with the theoretical power consumption and the attenuation compensation coefficient to calculate the actual power consumption.

[0012] Furthermore, the method for charging the host battery and the slave battery is as follows: A1: Arrange all energy storage batteries into a battery queue according to the battery remaining capacity from small to large, and record the energy storage battery at the first position in the battery queue as the host battery. A2: The energy storage battery located in the second position is designated as the slave battery. The voltage difference is calculated by subtracting the battery voltage of the master battery from the battery voltage of the slave battery. A3: Using the slave battery as the discharge target and the master battery as the charging target, control the slave battery to charge the master battery until the voltage difference between the slave battery and the master battery is less than or equal to the preset voltage difference threshold, and then stop and update the battery voltage and battery balance of the master battery. A4: Remove the second-ranked energy storage battery from the battery queue and repeat steps A2-A3 until all energy storage batteries in the battery queue have been traversed.

[0013] Furthermore, the dual parallel mechanism is as follows: the available battery capacity is greater than the lower limit of the capacity, and the discharge time of the available battery is less than the upper limit of the discharge time. The method for selecting available batteries is as follows: Energy storage batteries with a remaining capacity greater than the lower limit of the remaining capacity are recorded as candidate batteries, and the rated discharge power of the candidate batteries is retrieved. Divide the remaining capacity of the candidate battery by the rated discharge power to calculate the discharge duration, and record the candidate batteries whose discharge duration is less than the upper limit as usable batteries.

[0014] Furthermore, the method for synchronously supplying power to the load is as follows: The available power is calculated by subtracting the remaining power of each available battery from its lower limit. The actual power supply is then calculated by summing up all available power. When the actual power supply exceeds the actual power consumption, the energy storage inverter synchronously supplies power to the load from all available batteries and calculates the total power supply to the load until the total power supply equals the actual power consumption. When the actual power supply is equal to or less than the actual power consumption, the energy storage inverter synchronously supplies power to the load from all available batteries until all available power is exhausted.

[0015] The technical advantages of the control system for off-grid parallel connection of energy storage inverters in this invention are as follows: (1): This invention constructs a multidimensional trend curve by combining time granularity analysis and dynamic sliding window tracking, and introduces a joint analysis method of the tangent slope of the first derivative and the curve curvature of the second derivative. It can accurately capture the critical moment when the real-time state of the power grid changes significantly and irreversibly, effectively eliminate the interference factors caused by normal fluctuations in the power grid, and avoid the problem of false triggering or delayed response of off-grid parallel operation caused by a single threshold or simple fluctuation judgment. Thus, it provides a reliable time reference for subsequent calculation of off-grid duration and estimation of actual power consumption, ensures the accuracy of off-grid parallel control in timing logic, and improves the reliability of off-grid parallel operation.

[0016] (2): By introducing a dual parallel mechanism, this invention provides a dual threshold for the identification and control of available batteries, including a minimum power level and a minimum duration. This avoids safety risks and hidden dangers caused by over-discharge of batteries, and also avoids the problem of low overall power supply performance or frequent switching caused by inefficient batteries being connected to the parallel system. Thus, it extends the power supply time as much as possible with limited power resources, and provides a longer window for subsequent emergency handling, thereby achieving a matching effect between power supply capacity and load demand. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the architecture of a control system for an off-grid parallel connection of an energy storage inverter provided in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of a control system for an off-grid parallel connection of an energy storage inverter, provided in Embodiment 1 of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Please refer to Figures 1-2 As shown in this embodiment, a control system for off-grid parallel operation of an energy storage inverter is applied to an off-grid parallel platform and includes: The off-grid parallel connection determination module collects the online power supply characteristics of the power grid during the control period, predicts the power grid status in the next control period through the power grid status identification model, and determines whether the energy storage inverter enters the off-grid parallel connection mode based on the power grid status. The off-grid parallel control system is a system that controls the power supply of the energy storage battery in real time according to the specific operating status of the energy storage inverter. The off-grid parallel control system includes an off-grid parallel platform, the power grid, the load end, the energy storage battery, and the energy storage inverter; among them, the number of energy storage batteries is not unique, thus providing the hardware foundation for off-grid parallel operation.

[0020] An energy storage inverter is a power device that connects energy storage batteries to the power grid and load, and can switch between grid-connected and off-grid modes. Under normal grid conditions, the energy storage inverter is in grid-connected mode and operates in parallel with the grid to achieve bidirectional flow of power. When the grid experiences a fault, voltage drop, or frequency anomaly, the energy storage inverter needs to switch to off-grid mode to operate independently of the grid, thereby providing a continuous and stable power supply to the load.

[0021] Online power supply characteristics are a comprehensive representation of the power supply status of the power grid within a single control period. They can describe the power supply quality of the power grid from multiple dimensions, such as the power grid itself, equipment fluctuation status, and environmental factors. Online power supply characteristics include voltage sequence, frequency sequence, power balance sequence, and equipment status sequence.

[0022] When collecting online power supply characteristics, it is necessary to first determine the control period for parameter acquisition. This control period serves as the duration for one online power supply characteristic acquisition and also as the duration for subsequent off-grid parallel control of the energy storage inverter. When determining the control period, a periodic and continuously rolling logic is followed. That is, according to the underlying rule of uninterrupted communication in the power grid system, the overall operating time of the power grid system is divided into multiple intervals of equal duration. These intervals are connected end to end and continuously cover the entire time axis, thus forming multiple control periods that are continuously distributed along the timeline.

[0023] Voltage sequence refers to the voltage values ​​of each phase of the power grid at each time point during the control period, which can provide data support for the changes in power grid voltage; In this embodiment, when collecting voltage sequences, the A-phase voltage, B-phase voltage, and C-phase voltage of the power grid are periodically sampled within the control period using a preset sampling frequency as the standard, so as to obtain the A-phase voltage set, B-phase voltage set, and C-phase voltage set respectively. According to the order of acquisition time, the phase A voltage, phase B voltage, and phase C voltage are arranged into phase A voltage queue, phase B voltage queue, and phase C voltage queue, respectively. The phase A voltage queue, phase B voltage queue, and phase C voltage queue are then synchronized and merged to generate a voltage sequence.

[0024] Frequency sequence refers to the value of the power grid frequency at each time node within the control period, which can provide data support for the frequency changes of the power grid; In this embodiment, when collecting frequency sequences, the power grid frequency is periodically sampled within the control period using a preset sampling frequency as the standard to obtain a set of power grid frequencies, and the power grid frequencies are arranged into a frequency sequence according to the order of collection time.

[0025] The power balance sequence refers to the values ​​of various types of power at each time point during the control period of the power grid, which can provide data support for the power changes of the power grid. In this embodiment, when collecting the power balance sequence, the active power and reactive power of the power grid are periodically sampled within the control period using a preset sampling frequency as the standard, so as to obtain the active power set and the reactive power set respectively. According to the order of collection time, the active power and reactive power are arranged into active power queue and reactive power queue respectively. The active power queue and the reactive power queue are then synchronized and merged to generate a power balance sequence.

[0026] Equipment status sequence refers to the result of the operating status of power grid equipment at each time node within the control period, which can provide data support for changes in power grid equipment; In this embodiment, when collecting the device status sequence, the switching status and operating status of the power grid equipment are periodically sampled within the control period using a preset sampling frequency as the standard, to obtain a set of switching status and a set of operating status respectively; the switching status includes on and off; the operating status includes normal and abnormal. According to the order of collection time, the switch status and the running status are arranged into a switch status queue and a running status queue, respectively. The switch status queue and the running status queue are then synchronized and merged to generate a device status sequence.

[0027] It should be noted that the preset sampling frequency refers to the rate at which data is collected every certain interval when collecting power grid data. According to the Nyquist theorem, the sampling frequency must be greater than twice the highest frequency component of the collected signal. That is, when the standard frequency of the power grid is 50Hz, the preset sampling frequency must be greater than 100Hz.

[0028] After obtaining the online power supply characteristics, in order to fully explore the deep coupling relationship and complementary information between the parameters of each dimension, it is necessary to perform multimodal feature extraction and fusion on the online power supply characteristics, so as to generate a multimodal fusion vector for subsequent model recognition. Specifically, the method for generating multimodal fusion vectors is as follows: Time-domain features and frequency-domain features are extracted from voltage sequence, frequency sequence, power balance sequence, and equipment status sequence, respectively. The time-domain features include mean, variance, and peak value, while the frequency-domain features include fundamental amplitude, harmonic distortion rate, and spectral energy distribution. The time-domain and frequency-domain features of the voltage sequence, frequency sequence, power balance sequence, and equipment state sequence are concatenated sequentially to obtain the local feature vectors of the voltage sequence, frequency sequence, power balance sequence, and equipment state sequence. A multi-head self-attention mechanism is used to perform cross-modal feature fusion on four local feature vectors to generate a fused multimodal fusion vector.

[0029] Cross-modal feature fusion through multi-head self-attention mechanism can adaptively learn the correlation weights between different modal features, so that the fused vector can more comprehensively reflect the operating status of the power grid and provide higher quality feature input for subsequent state prediction.

[0030] The power grid state identification model is a time-series prediction model built on a long short-term memory network. It takes a multimodal fusion vector as input data and the power grid state of the next control period corresponding to the current control period as output data, so that the power grid state of the next control period after the current control period can be identified in advance. The power grid status is a result of the power grid's operation during the control period. Specifically, the power grid status includes online status and offline status.

[0031] The training method for the power grid condition identification model is as follows: The online power supply characteristics of multiple historical control periods are extracted from the historical database, as well as the power grid status of the next control period corresponding to the multiple historical control periods. The power grid status is numbered and converted, with online status converted to 11 and offline status converted to 00. Multimodal feature extraction and fusion are performed on the online power supply characteristics to obtain a multimodal fusion vector. The multimodal fusion vectors are arranged according to the acquisition time, and the prediction time step, sliding step, and sliding window length are preset. The training dataset is constructed by using the multimodal fusion feature vector as input data and the power grid state of the next control period after the predicted time step as output data. The training dataset is input into the Long Short-Term Memory network for iterative training until the loss function converges to the preset loss threshold, thus obtaining the power grid state recognition model.

[0032] In this embodiment, the online power supply characteristics of the current control period are fused into a multimodal fusion vector and input into the power grid state recognition model. When the output data is 11, the power grid state of the next control period is online; when the output data is 00, the power grid state of the next control period is offline.

[0033] It should be noted that when the grid status of the next control period is offline, it does not mean that the entire next control period is offline. Rather, it means that the first part of the next control period is online and the second part is offline, or that the entire period is offline. After obtaining the grid status of the next control period, it is possible to determine whether to control the energy storage inverter to enter the off-grid parallel mode. In this embodiment, when determining whether the energy storage inverter has entered the off-grid parallel mode, if the grid is online, the energy storage inverter is working normally, and it is determined that the energy storage inverter has not entered the off-grid parallel mode; if the grid is offline, the energy storage inverter is not working normally, and it is determined that the energy storage inverter has entered the off-grid parallel mode.

[0034] The subsequent off-grid parallel control process will be triggered only after the energy storage inverter enters the off-grid parallel mode, thus providing a logical basis for subsequent operations.

[0035] The off-grid time determination module performs time-granularity analysis and dynamic sliding window tracking on the online power supply characteristics, plots the trend curve, extracts the geometric features of the trend curve, and determines the off-grid time based on the geometric features. After the energy storage inverter enters the off-grid parallel mode, it is necessary to determine the critical moment when the grid performs the off-grid parallel operation, so as to provide an accurate time reference for subsequent power consumption calculation and power supply strategy formulation.

[0036] Time granularity analysis refines the time resolution of online power supply characteristics within the control period to each specific moment, determining the precise value of the online power supply characteristics at each moment; Dynamic sliding window tracking refers to using a fixed-length sliding window to continuously slide on the time axis to perform real-time analysis and feature extraction of online power supply characteristics within the control period, in order to capture the dynamic changing trend of online power supply characteristics, and to continuously represent the dynamic changing trend by drawing a trend curve. The trend curves include voltage curves, frequency curves, power curves, and equipment curves.

[0037] By performing time-granularity analysis and dynamic sliding window tracking on the online power supply characteristics, the displayed values ​​of the online power supply characteristics at each moment can be obtained, namely: the displayed values ​​of voltage, frequency, power balance, and equipment status at each moment.

[0038] The method for drawing the trend curve is as follows: Using the displayed values ​​of voltage, frequency, power balance, and equipment status at all times as the vertical axis and time as the horizontal axis, the corresponding coordinate points are marked in a two-dimensional coordinate system to generate voltage coordinate system, frequency coordinate system, power coordinate system, and equipment coordinate system. In the voltage coordinate system, frequency coordinate system, power coordinate system, and equipment coordinate system, coordinate points at adjacent positions are connected sequentially by smooth curves to draw voltage curves, frequency curves, power curves, and equipment curves.

[0039] The trend curve is a continuous curve showing the change of the power supply characteristics of the power grid over time. It is used to accurately determine the critical moment when the power grid transitions from an online state to an offline state. In order to accurately extract the offline moment from the trend curve, it is necessary to extract the geometric features of the trend curve. Geometric features include slope abrupt change points and curvature abrupt change points; these are used to represent points where the slope of the curve changes significantly and points where the curvature of the curve changes significantly, respectively.

[0040] The method for extracting geometric features is as follows: In the voltage curve, frequency curve, power curve, and equipment curve, the slope of the tangent is calculated by subtracting the displayed value of the next coordinate point from the displayed value of the previous coordinate point and dividing by the displayed value of the previous coordinate point. Following the chronological order, all tangent slopes are aggregated into a slope sequence. Using a preset window length as a standard, the slope sequence is divided into A continuously distributed slope windows. The preset window length refers to a fixed time length set when analyzing the data in the sequence. Within this time length, several consecutive data points need to be analyzed as a whole. Specifically, the preset window length can be customized according to actual needs; for example, the preset window length is the duration corresponding to 10 consecutive time points. Within A slope windows, the slope difference is calculated by subtracting the slope of the last tangent from the slope of the first tangent. The coordinates of points where the slope difference is greater than a preset slope difference threshold are recorded as slope abrupt change points. The preset slope difference threshold refers to the minimum slope difference value corresponding to the coordinates of the points recorded as slope abrupt change points, thus providing a numerical basis for identifying slope abrupt change points. Specifically, the preset slope difference threshold is obtained by statistically analyzing the minimum slope difference values ​​corresponding to a large number of historical coordinates of points recorded as slope abrupt change points and then averaging them. Starting from the second coordinate point, the three coordinate points in adjacent positions are denoted as a point set, and the curvature of the curve of the point set is calculated one by one using the curvature formula; According to the chronological order, all curve curvatures are summarized into a curvature sequence. Based on the preset window duration, the curvature sequence is divided into B continuously distributed curvature windows. Within B curvature windows, the curvature difference is calculated by subtracting the curvature of the last curve from that of the first curve. Points with curvature differences greater than a preset curvature difference threshold are designated as mutation sets, and the coordinate point at the center of this mutation set is designated as a curvature mutation point. The preset curvature difference threshold refers to the minimum curvature difference value corresponding to the coordinate points designated as curvature mutation points, thus providing a numerical basis for identifying curvature mutation points. Specifically, the preset curvature difference threshold is obtained by statistically averaging the minimum curvature difference values ​​corresponding to a large number of historically designated curvature mutation points.

[0041] It should be noted that the slope abrupt change point reflects the moment when the rate of change of online power supply characteristics changes significantly, and these moments often correspond to the turning point of the power grid's operating state; the curvature abrupt change point reflects the moment when the degree of curvature of the trend of change of online power supply characteristics changes significantly, and these moments also correspond to the turning point of the power grid's operating state.

[0042] The method for determining the off-network time is as follows: In voltage curves, frequency curves, power curves, and equipment curves, rated voltage lines and rated frequency lines are plotted respectively, using rated voltage and rated frequency as standards; rated voltage refers to the standard voltage under normal power grid conditions, and rated frequency refers to the standard frequency under normal power grid conditions. Along the vertical axis, calculate the first vertical distance between the slope abrupt change point and the rated voltage line, and between the slope abrupt change point and the rated frequency line, and record the slope abrupt change point corresponding to the minimum value of the first vertical distance as a type of abrupt change point. Along the vertical axis, calculate the second longitudinal distance between the curvature abrupt change point and the rated voltage line, and between the curvature abrupt change point and the rated frequency line, and record the curvature abrupt change point corresponding to the minimum value of the second longitudinal distance as the second type of abrupt change point; Arrange all the Class I and Class II mutation points into a mutation queue according to the chronological order. The time corresponding to the last Class I or Class II mutation point in the mutation queue is recorded as the off-network time.

[0043] By combining time granularity analysis, dynamic sliding window tracking, and slope and curvature changes in analysis, the critical moment when the power grid transitions from an online to an offline state can be accurately captured from the geometric characteristics of the trend curve, thus providing a precise time reference for subsequent electricity consumption calculations and power supply strategy formulation.

[0044] The power consumption calculation module collects the load power consumption parameters of the load end during the control period, predicts the theoretical power consumption of the load end in the next control period through the power consumption prediction model, determines the off-grid duration based on the off-grid time, and calculates the actual power consumption by combining the off-grid duration and the theoretical power consumption. After the energy storage inverter switches to off-grid parallel mode, the energy storage battery needs to provide a continuous power supply to the load. Since the load's power demand is affected by various factors such as weather conditions, power consumption time, and load power changes, it is necessary to accurately predict the load's power consumption in the next control period in order to formulate a precise power supply strategy.

[0045] Load power consumption parameters are a comprehensive representation of the power consumption status of the load end within a single control period. They can describe the power demand of the load end from multiple dimensions such as weather environment, power consumption period and load power changes. Load electrical parameters include temperature, humidity, light intensity, time of day, and load power.

[0046] Temperature, humidity, and light intensity represent the weather environment at the load end, which is obtained through corresponding sensors.

[0047] Electricity consumption periods are a representation of the specific time period type of the load end during the control period. Specifically, electricity consumption periods include holiday peak, holiday valley peak, holiday off-peak, weekday peak, weekday valley peak, and weekday off-peak. In this embodiment, after obtaining the electricity consumption periods, it is necessary to convert the electricity consumption periods into numbers. For example, holiday peak is converted to 1-GG, holiday valley peak is converted to 1-DG, holiday off-peak is converted to 1-ZP, weekday peak is converted to 0-GG, weekday valley peak is converted to 0-DG, and weekday off-peak is converted to 0-ZP.

[0048] Load power refers to the fluctuation of load power at the load end during the control period, reflecting the changes in the start-up, shutdown and operating status of electrical equipment; load power is obtained through power sensor queries.

[0049] The electricity consumption prediction model is a time-series prediction model built on a long short-term memory network. It takes load electricity consumption parameters as input data and the theoretical electricity consumption of the load in the next control period as output data, so that the theoretical electricity consumption of the load in the next control period after the current control period can be identified in advance.

[0050] It should be noted that the training logic and training method of the electricity consumption prediction model are consistent with those of the above-mentioned power grid state identification model; the difference between the two is that the input data of the electricity consumption prediction model is the load electricity consumption parameters, and the output data is the theoretical electricity consumption.

[0051] In this embodiment, the load power consumption parameters of the current control period are input into the power consumption prediction model, and the theoretical power consumption of the load in the next control period is output.

[0052] The theoretical power consumption is the power consumption of the load end during the entire period of the next control period, so that the theoretical power consumption corresponds to the overall duration of the control period. Since the off-grid time of the power grid in the next control period is not fixed, before the off-grid time, the power consumption of the load end is supplied by the power grid in real time, and after the off-grid time, the power consumption of the load end is supplied by the energy storage battery. Therefore, it is necessary to calculate the power consumption after the off-grid time in the next control period to obtain the required actual power consumption. The actual electricity consumption is calculated as follows: The current time is taken as the start time of the next control period, and the duration between the start time and the offline time is recorded as the online duration; The offline time is calculated by subtracting the duration of the controlled period from the online time. The actual power consumption is calculated by comparing the off-grid duration with the control period duration and combining it with the theoretical power consumption and the attenuation compensation coefficient. The formula for calculating actual electricity consumption is: ; In the formula, This represents the actual electricity consumption. Theoretical electricity consumption For offline time, To control the duration of the time slot, This is the attenuation compensation coefficient.

[0053] It should be noted that the attenuation compensation coefficient is a coefficient used to compensate for the attenuation of electricity consumption caused by factors such as line wear and equipment wear during the control period. This can overcome the error in electricity consumption calculation caused by negative attenuation in different dimensions of the load, and ensure that the actual electricity consumption can better match the actual electricity demand of the load. The attenuation compensation coefficient is calculated based on historical data; specifically, the value range of the attenuation compensation coefficient is 1.05-1.21.

[0054] The battery charging optimization module obtains the remaining battery capacity and battery voltage of the energy storage battery, divides the energy storage battery into master battery and slave battery according to the remaining battery capacity, and optimizes the charging of master battery and slave battery according to the battery voltage. In off-grid parallel mode, multiple energy storage batteries need to be connected in parallel to supply power to the load. Due to differences in factory parameters, usage history, and aging levels among the various energy storage batteries, the state of charge and terminal voltage of the energy storage batteries are often inconsistent. If energy storage batteries with large voltage differences are directly connected in parallel to supply power, the lower voltage energy storage battery will be charged by the higher voltage energy storage battery, resulting in energy loss and reduced charging efficiency. In severe cases, it may also lead to over-discharge of some batteries and underutilization of the remaining capacity of some batteries. Therefore, it is necessary to optimize the voltage balancing of each energy storage battery before connecting them in parallel to supply power.

[0055] Battery capacity refers to the remaining charge of the energy storage battery at the current moment, and battery voltage refers to the terminal voltage of the energy storage battery at the current moment. In this embodiment, both battery capacity and battery voltage are obtained by querying the battery management system.

[0056] Voltage difference refers to the difference between the battery voltages of two energy storage batteries, which can be used as a numerical basis for charging energy storage batteries based on their battery voltage levels.

[0057] The methods for charging the host battery and slave battery are as follows: A1: Arrange all energy storage batteries into a battery queue according to the battery remaining capacity from small to large, and record the energy storage battery at the first position in the battery queue as the host battery. A2: The energy storage battery located in the second position is designated as the slave battery. The voltage difference is calculated by subtracting the battery voltage of the master battery from the battery voltage of the slave battery. A3: Using the slave battery as the discharge target and the master battery as the charging target, control the slave battery to charge the master battery until the voltage difference between the slave battery and the master battery is less than or equal to the preset voltage difference threshold, and then stop and update the battery voltage and battery balance of the master battery. A4: Remove the second-ranked energy storage battery from the battery queue and repeat steps A2-A3 until all energy storage batteries in the battery queue have been traversed, thus achieving optimized charging of the energy storage batteries.

[0058] It should be noted that when the slave battery optimizes the charging of the master battery, the voltage of the slave battery after the charging optimization will not be lower than the voltage of the master battery.

[0059] The preset differential voltage threshold refers to the maximum allowable difference in battery voltage between each energy storage battery. For example, the preset differential voltage threshold can be set to 0.5V. When the voltage difference of all energy storage batteries is kept within the preset differential voltage range, the battery voltage of each energy storage battery is basically consistent, and parallel power supply can be safely performed.

[0060] The off-grid parallel battery module, based on a dual parallel mechanism, selects available batteries from the energy storage batteries, and after connecting the available batteries off-grid in parallel, synchronously supplies power to the load. After the energy storage batteries have completed the voltage balancing optimization, it is necessary to screen all energy storage batteries for availability and select those that can meet the off-grid power supply requirements of the load.

[0061] When screening available batteries, it is necessary to do so under the constraint of a dual parallel mechanism to ensure that the selected available batteries can maintain a reasonable balance in terms of both power supply capacity and power supply duration. Specifically, the dual parallel mechanism is as follows: the available battery capacity is greater than the lower limit of the capacity, and the discharge time of the available battery is less than the upper limit of the discharge time.

[0062] The method for selecting available batteries is as follows: The remaining capacity of the energy storage battery is compared with the lower limit of the remaining capacity, and the energy storage battery with a remaining capacity greater than the lower limit of the remaining capacity is recorded as a candidate battery. Find the rated discharge power of the candidate battery, divide the remaining battery capacity of the candidate battery by the rated discharge power, and calculate the discharge time. The discharge duration of candidate batteries is compared with the upper limit of the duration, and candidate batteries whose discharge duration is less than the upper limit of the duration are marked as usable batteries.

[0063] In this embodiment, the lower limit of the reserve capacity refers to the minimum battery reserve requirement for the energy storage battery to be recognized as a usable battery; for example, the lower limit of the reserve capacity can be set to 10% of the rated capacity of the energy storage battery; the upper limit of the duration refers to the maximum charging duration requirement for the energy storage battery to be recognized as a usable battery; specifically, the upper limit of the duration is directly proportional to the off-grid duration, that is: the longer the off-grid duration, the larger the upper limit of the duration; for example, the upper limit of the duration is 30% of the off-grid duration.

[0064] After the available batteries are selected, all available batteries can be connected in parallel off-grid, and multiple independent available batteries can be connected in parallel into a whole structure, so that all available batteries can supply power to the load synchronously as a whole. The method for synchronously supplying power to the load is as follows: The available power is calculated by subtracting the remaining power of each available battery from its lower limit. The actual power supply is then calculated by summing up all available power supplies and comparing the actual power supply with the actual power consumption. When the actual power supply exceeds the actual power consumption, the energy storage inverter synchronously supplies power to the load from all available batteries and calculates the total power supply to the load until the total power supply equals the actual power consumption. When the actual power supply is equal to or less than the actual power consumption, the energy storage inverter synchronously supplies power to the load from all available batteries until all available power is exhausted.

[0065] In this embodiment, if the actual power supply is less than the actual power consumption, it means that the available power in the available batteries cannot meet the power demand of the load in the next control period. At this time, even if some of the battery capacity in the energy storage batteries is not utilized, the remaining battery capacity of the energy storage batteries will not be utilized. This not only ensures that the available batteries remain in a safe and reliable state after the power supply ends, but also avoids the participation of energy storage batteries that cannot meet the power demand of the load in the battery capacity scheduling operation. This can give full play to the collaborative power supply capability of multiple available batteries, thereby improving the reliability and continuity of the available batteries' power supply operation to the load.

[0066] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A control system for off-grid parallel operation of an energy storage inverter, applied to an off-grid parallel platform, characterized in that, include: The off-grid parallel connection determination module is used to collect the online power supply characteristics of the power grid during the control period, predict the power grid status in the next control period through the power grid status identification model, and determine whether the energy storage inverter enters the off-grid parallel connection mode based on the power grid status. The off-grid time determination module is used to perform time-granularity analysis and dynamic sliding window tracking of online power supply characteristics, draw trend curves, extract geometric features of trend curves, and determine the off-grid time of the power grid based on the geometric features. The power consumption calculation module is used to collect the load power consumption parameters of the load end during the control period, predict the theoretical power consumption of the load end in the next control period through the power consumption prediction model, determine the off-grid duration based on the off-grid time, and calculate the actual power consumption by combining the off-grid duration and the theoretical power consumption. The battery charging optimization module is used to obtain the remaining battery capacity and battery voltage of the energy storage battery, divide the energy storage battery into master battery and slave battery according to the remaining battery capacity, and optimize the charging of master battery and slave battery according to the battery voltage. The off-grid parallel battery module is used to select available batteries from energy storage batteries based on a dual parallel mechanism, and then connect the available batteries off-grid in parallel to synchronously supply power to the load.

2. The control system for off-grid parallel connection of an energy storage inverter according to claim 1, characterized in that, Online power supply characteristics include voltage sequence, frequency sequence, power balance sequence, and device status sequence; The power grid status includes online status and offline status. The online status is converted to 11, and the offline status is converted to 00. When the output of the power grid status identification model is 11, the power grid status is online; when the output of the power grid status identification model is 00, the power grid status is offline.

3. The control system for off-grid parallel connection of an energy storage inverter according to claim 2, characterized in that, When the grid is online, the energy storage inverter is determined not to enter the off-grid parallel mode; when the grid is offline, the energy storage inverter is determined to enter the off-grid parallel mode.

4. The control system for off-grid parallel connection of an energy storage inverter according to claim 3, characterized in that, The trend curves include voltage curves, frequency curves, power curves, and equipment curves; The method for drawing the trend curve is as follows: Using the displayed values ​​of voltage, frequency, power balance, and equipment status at all times as the vertical axis and time as the horizontal axis, coordinate points are marked in a two-dimensional coordinate system to generate voltage coordinate system, frequency coordinate system, power coordinate system, and equipment coordinate system. In the voltage coordinate system, frequency coordinate system, power coordinate system, and equipment coordinate system, two coordinate points at adjacent positions are connected sequentially by a smooth curve to draw the voltage curve, frequency curve, power curve, and equipment curve.

5. The control system for off-grid parallel connection of an energy storage inverter according to claim 4, characterized in that, Geometric features include points of abrupt change in slope and points of abrupt change in curvature; The method for extracting geometric features is as follows: In the voltage curve, frequency curve, power curve, and equipment curve, the slope of the tangent is calculated by subtracting the displayed value of the next coordinate point from the displayed value of the previous coordinate point and dividing by the displayed value of the previous coordinate point. According to the chronological order, all tangent slopes are summarized into a slope sequence, and the slope sequence is divided into continuously distributed slope windows based on the preset window duration. In the slope window, the slope difference is calculated by subtracting the slope of the last tangent from the slope of the first tangent, and the coordinates of the points where the slope difference is greater than the preset slope difference threshold are recorded as slope change points. Starting from the second coordinate point, the three coordinate points in adjacent positions are denoted as a point set, and the curvature of the curve of the point set is calculated using the curvature formula; According to the chronological order, all curve curvatures are summarized into a curvature sequence, and the curvature sequence is divided into continuously distributed curvature windows based on the preset window duration. In the curvature window, the curvature difference is calculated by subtracting the curvature of the last curve from the curvature of the first curve. The set of points whose curvature difference is greater than the preset curvature difference threshold is recorded as the mutation set, and the coordinate point in the middle of the mutation set is recorded as the curvature mutation point.

6. The control system for off-grid parallel connection of an energy storage inverter according to claim 5, characterized in that, The method for determining the off-network time is as follows: In the voltage curve, frequency curve, power curve, and equipment curve, the rated voltage line and rated frequency line are plotted respectively, using the rated voltage and rated frequency as the standard. Along the vertical axis, calculate the first vertical distance between the slope abrupt change point and the rated voltage line, and between the slope abrupt change point and the rated frequency line, and record the slope abrupt change point corresponding to the minimum value of the first vertical distance as a type of abrupt change point. Along the vertical axis, calculate the second longitudinal distance between the curvature abrupt change point and the rated voltage line, and between the curvature abrupt change point and the rated frequency line, and record the curvature abrupt change point corresponding to the minimum value of the second longitudinal distance as the second type of abrupt change point; Arrange all the Class I and Class II mutation points into a mutation queue according to the chronological order. The time corresponding to the last Class I or Class II mutation point in the mutation queue is recorded as the off-network time.

7. The control system for off-grid parallel connection of an energy storage inverter according to claim 6, characterized in that, Load electrical parameters include temperature, humidity, light intensity, time of day, and load power; The actual electricity consumption is calculated as follows: The current time is taken as the start time of the next control period, and the duration between the start time and the offline time is recorded as the online duration; The offline time is calculated by subtracting the duration of the control period from the online time. The offline time is then compared with the duration of the control period and combined with the theoretical power consumption and the attenuation compensation coefficient to calculate the actual power consumption.

8. The control system for off-grid parallel connection of an energy storage inverter according to claim 7, characterized in that, The methods for charging the host battery and slave battery are as follows: A1: Arrange all energy storage batteries into a battery queue according to the battery remaining capacity from smallest to largest, and record the energy storage battery at the first position in the battery queue as the host battery. A2: The energy storage battery located in the second position is designated as the slave battery. The voltage difference is calculated by subtracting the battery voltage of the master battery from the battery voltage of the slave battery. A3: Using the slave battery as the discharge target and the master battery as the charging target, control the slave battery to charge the master battery until the voltage difference between the slave battery and the master battery is less than or equal to the preset voltage difference threshold, and then stop and update the battery voltage and battery balance of the master battery. A4: Remove the second-ranked energy storage battery from the battery queue and repeat steps A2-A3 until all energy storage batteries in the battery queue have been traversed.

9. A control system for off-grid parallel connection of an energy storage inverter according to claim 8, characterized in that, The dual parallel mechanism is as follows: the available battery capacity is greater than the lower limit of the capacity, and the discharge time of the available battery is less than the upper limit of the discharge time. The method for selecting available batteries is as follows: Energy storage batteries with a remaining capacity greater than the lower limit of the remaining capacity are recorded as candidate batteries, and the rated discharge power of the candidate batteries is retrieved. Divide the remaining capacity of the candidate battery by the rated discharge power to calculate the discharge duration, and record the candidate batteries whose discharge duration is less than the upper limit as usable batteries.

10. A control system for off-grid parallel connection of an energy storage inverter according to claim 9, characterized in that, The method for synchronously supplying power to the load is as follows: The available power is calculated by subtracting the remaining power of each available battery from its lower limit. The actual power supply is then calculated by summing up all available power. When the actual power supply exceeds the actual power consumption, the energy storage inverter synchronously supplies power to the load from all available batteries and calculates the total power supply to the load until the total power supply equals the actual power consumption. When the actual power supply is equal to or less than the actual power consumption, the energy storage inverter synchronously supplies power to the load from all available batteries until all available power is exhausted.

Images