Intelligent energy storage control method, device and equipment based on bidirectional energy management

Abstract

The invention relates to the technical field of intelligent energy storage control, and discloses an intelligent energy storage control method, device and equipment based on bidirectional energy management. The method comprises the following steps: carrying out real-time acquisition and power grid state analysis on power grid parameters, energy storage module parameters, load requirements and photovoltaic power generation data to obtain power grid state evaluation indexes and a system operation mode division result; executing a multi-objective optimization algorithm to obtain an energy scheduling plan including a charging and discharging power instruction; converting the energy scheduling plan into charging and discharging current control parameters, and inputting the charging and discharging current control parameters into a control structure composed of a prediction control layer and a real-time response control layer in the three-phase hybrid inverter to obtain power output data; the power output data of the multiple three-phase hybrid inverters are uploaded to the intelligent management cloud platform, multi-machine parallel coordination control is executed, a distributed energy network management scheme is obtained, the dynamic characteristics of the system under different electric energy flow directions are predicted, and the stability and adaptability of the distributed energy network are enhanced.

Description

Technical Field

[0001] The present invention relates to the field of intelligent energy storage control technology, and in particular to an intelligent energy storage control method, device and equipment based on bidirectional energy management. Background Art

[0002] Distributed energy networks consisting of solar power generation systems and intelligent energy storage systems have become one of the ideal topologies to meet the electricity needs of modern households and businesses. However, this photovoltaic-grid-energy storage hybrid architecture faces many technical challenges in practical applications, including power supply instability caused by the mismatch between grid fluctuations and user electricity demand, power fluctuation management during peak and trough periods, and how to efficiently coordinate the conversion and scheduling of multiple energy forms, all of which seriously affect the overall performance of the system and user experience.

[0003] The rapid growth of modern household and commercial electricity consumption has led to rising electricity bills and peak electricity demand. Traditional energy management systems lack effective grid state perception and prediction capabilities and are unable to flexibly adjust charging and discharging strategies based on grid state and electricity price changes. At the same time, existing energy storage control methods often use simplified system models, ignoring the dynamic characteristics of power flow and the complex interactions between system components, making it difficult to achieve accurate power distribution and smooth energy conversion, especially when the grid is subject to large disturbances or rapid load changes. Summary of the invention

[0004] The present invention provides an intelligent energy storage control method, device and equipment based on bidirectional energy management. The present invention predicts the dynamic characteristics of the system under different electric energy flows and enhances the stability and adaptability of the distributed energy network.

[0005] In a first aspect, the present invention provides an intelligent energy storage control method based on bidirectional energy management, the intelligent energy storage control method based on bidirectional energy management comprising: Real-time collection and grid status analysis of grid parameters, energy storage module parameters, load demand and photovoltaic power generation data are carried out to obtain grid status evaluation indicators and system operation mode classification results; Based on the grid state evaluation index and the system operation mode division result, a multi-objective optimization algorithm is executed to obtain an energy scheduling plan including charging and discharging power instructions; The energy scheduling plan is converted into charge and discharge current control parameters, and input into a control structure composed of a prediction control layer and a real-time response control layer in a three-phase hybrid inverter to obtain power output data; The power output data of multiple three-phase hybrid inverters are uploaded to the intelligent management cloud platform, and multi-machine parallel coordinated control is performed to obtain a distributed energy network management solution.

[0006] In a second aspect, the present invention provides an intelligent energy storage control device based on bidirectional energy management, the intelligent energy storage control device based on bidirectional energy management comprising: Real-time acquisition module, used for real-time acquisition of grid parameters, energy storage module parameters, load demand and photovoltaic power generation data and grid status analysis, to obtain grid status evaluation indicators and system operation mode classification results; A multi-objective optimization module, used to execute a multi-objective optimization algorithm based on the grid state evaluation index and the system operation mode division result to obtain an energy scheduling plan including charging and discharging power instructions; A prediction control module, used for converting the energy scheduling plan into charge and discharge current control parameters, and inputting the parameters into a control structure composed of a prediction control layer and a real-time response control layer in a three-phase hybrid inverter to obtain power output data; The parallel coordination control module is used to upload the power output data of multiple three-phase hybrid inverters to the intelligent management cloud platform, perform multi-machine parallel coordination control, and obtain a distributed energy network management solution.

[0007] The third aspect of the present invention provides an intelligent energy storage control device based on bidirectional energy management, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the intelligent energy storage control device based on bidirectional energy management executes the above-mentioned intelligent energy storage control method based on bidirectional energy management.

[0008] In the technical solution provided by the present invention, a high-precision, multi-frequency parameter real-time acquisition system is used to realize comprehensive monitoring of the power grid state, energy storage state, load demand and photovoltaic power generation, providing a reliable data basis for system operation and making the energy flow identification more accurate. The power grid state evaluation mechanism based on power grid harmonic content analysis and weighted average composite function, combined with the electricity price interval division technology of fuzzy clustering algorithm, enables the system to intelligently perceive the changes in power grid state and automatically adjust the operation mode according to the characteristics of electricity prices. The parameterized representation method of power flow is introduced to express the system matrix and input matrix as functions of three states: charging, discharging and standby, which greatly simplifies the controller design and accurately describes the dynamic characteristics of the system under different power flows. A multi-objective optimization algorithm is used for energy scheduling, comprehensively considering multiple objectives such as electricity cost, battery life, self-generation and self-use rate, and power grid interaction support, and dynamically adjusting the weight coefficient according to the system operation mode to achieve the optimal configuration of resources. A two-level control structure combining the predictive control layer and the real-time response control layer was designed. With the smooth transition strategy and power smoothing algorithm, the system oscillation and grid interaction power fluctuations were effectively suppressed, ensuring the stable operation of the three-phase hybrid inverter under various working conditions. The remote monitoring of system operation data and the coordinated control of multiple machines in parallel were realized through the intelligent management cloud platform. The master-slave droop control algorithm and virtual synchronous machine technology were adopted to enhance the stability and adaptability of the distributed energy network, especially the autonomous operation capability when communication was interrupted. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0010] Figure 1 A schematic diagram of a flow chart of an intelligent energy storage control method based on bidirectional energy management provided in an embodiment of the present application; Figure 2 A schematic block diagram of the structure of an intelligent energy storage control device based on bidirectional energy management provided in an embodiment of the present application; Figure 3 A schematic block diagram of the structure of an intelligent energy storage control device based on bidirectional energy management provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0012] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change based on actual conditions.

[0013] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0014] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0015] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0016] See also Figure 1 , Figure 1 A flow chart of an intelligent energy storage control method based on bidirectional energy management provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the intelligent energy storage control method based on bidirectional energy management provided in the embodiment of the present application includes steps S100 to S400.

[0017] Step S100: Real-time collection and grid status analysis of grid parameters, energy storage module parameters, load demand and photovoltaic power generation data are performed to obtain grid status evaluation indicators and system operation mode classification results; It is understandable that the execution subject of the present invention may be an intelligent energy storage control device based on bidirectional energy management, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0018] Specifically, the voltage, frequency, phase and electricity price information of the three-phase power grid are sampled with high precision to ensure that the measurement results of the power grid parameters are sufficiently accurate and real-time. The sampling process uses a high-frequency data acquisition module with a sampling frequency set to 10kHz to capture the instantaneous changes of the power grid and transmit the obtained power grid parameter data to the data processing unit for storage and subsequent analysis. At the same time, the state of charge, DC bus voltage, charge and discharge current and battery temperature of the battery pack are continuously monitored, among which the battery state of charge SOC is an important parameter for measuring the remaining capacity of the battery, and the DC bus voltage is used to judge the energy exchange between the battery and the system. The measurement of charge and discharge current helps to calculate the power input and output of the battery, while the monitoring of battery temperature can effectively prevent safety hazards caused by overheating. The user's load power consumption is continuously monitored to obtain complete load demand information. In this process, the power analyzer is used to measure the power consumption of the user's three-phase or single-phase load, and identify the peak period of the load and its change pattern. The sampling frequency is set to once per minute to ensure the continuity and accuracy of the data. At the same time, the photovoltaic power generation monitoring module synchronously measures the output voltage, current, power and other electrical parameters of the photovoltaic array, as well as external influencing factors such as environmental irradiance and temperature, and obtains high-precision photovoltaic power generation data at a sampling frequency of 100Hz. These data can reflect the real-time power output of photovoltaic power generation, and are combined with environmental factors to evaluate future power generation capacity. All collected grid parameters, energy storage module parameters, load demand data and photovoltaic power generation data are uniformly transmitted to the data processing unit, which relies on the ARM Cortex-A53 processor to perform data denoising, outlier detection and missing value interpolation and other preprocessing tasks. In the denoising link, the weighted moving average filtering method is used to remove high-frequency noise interference, and the median filtering algorithm is used to eliminate mutation anomalies. In the process of outlier detection, a threshold judgment model is constructed based on the distribution of historical data. If the measured value exceeds the set range, it is corrected by the interpolation algorithm to ensure the continuity and accuracy of the data. For missing values, linear regression and Kalman filtering methods are used to complete them to ensure data integrity. The preprocessed data will be stored in the local cache and archived in a unified data format, and timestamps will be marked for subsequent analysis. Based on the preprocessed data, the real-time values ​​of photovoltaic power generation, load demand power, battery charging and discharging power, and grid exchange power are compared to identify the direction of power flow and form a system parameter data set. In this process, an energy flow identification model is established, and the energy flow is divided into five basic modes, namely photovoltaic direct power supply mode, photovoltaic to battery charging mode, battery to load power supply mode, grid to battery charging mode, and battery to grid power supply mode.The calculation is performed based on the real-time measured photovoltaic power generation, user load power, battery charging and discharging power, and grid exchange power, and the current power flow direction is determined by the power balance relationship. If the photovoltaic power generation is greater than the load demand power and the remaining part is used to charge the battery, it is determined that the photovoltaic power generation mode is currently in the battery charging mode; if the battery power generation power is greater than the load demand power, it is determined to be in the battery power transmission mode. Through real-time analysis, the power flow direction information is dynamically updated and a system parameter data set is formed. The grid state is analyzed based on the system parameter data set to obtain the grid state evaluation index and the system operation mode division results.

[0019] The three-phase voltage amplitude, phase difference, frequency and fluctuation rate of the power grid in the system parameter data set are continuously monitored to reflect the dynamic changes of the power grid operation status. The data acquisition module collects the basic parameters of the power grid in real time with high precision and high sampling rate, and obtains the monitoring results of the basic parameters of the power grid by calculating the voltage amplitude change rate, phase deviation and frequency fluctuation between adjacent sampling points. Fourier transform is performed on the monitoring results of the basic parameters of the power grid to extract the harmonic components in the power grid signal, and the amplitude and total harmonic distortion rate THD of each order of harmonics are calculated to obtain the analysis results of the harmonic content of the power grid. Based on the analysis results of the harmonic content of the power grid, a weighted average composite function calculation method is used to comprehensively evaluate the power grid state and determine the evaluation index of the power grid state. This calculation method constructs a power grid state evaluation model by weighted summing multiple key factors such as power grid frequency stability, phase consistency, voltage fluctuation rate and harmonic distortion. In this model, the weight coefficient of each parameter is obtained based on historical operation data training to ensure that the evaluation index can accurately reflect the actual situation of the power grid state. The numerical range of the grid state evaluation index calculated by the system is set between 0 and 100. When the index value is higher than 85, it means that the grid is in a stable state; when the index value is between 60 and 85, it indicates that the grid is in a fluctuating state; and when the index value is lower than 60, the grid state is judged to be abnormal. Through this calculation process, the grid state is accurately evaluated, and the stability level of the grid is determined based on the evaluation index to obtain the grid state judgment result. At the same time, combined with electricity price information, to ensure that the division of operation modes optimizes energy management while reducing economic costs. After obtaining the real-time electricity price data in the system parameter data set, the fuzzy clustering algorithm is used to intelligently divide the electricity price and obtain the classification results of different electricity price ranges. The fuzzy clustering algorithm analyzes the time series variation characteristics of electricity prices and combines the regional electricity price policy to dynamically adjust the cluster center to ensure the rationality of electricity price division. Three main electricity price ranges are set, among which the electricity price in the peak range is higher than 130% of the daily average electricity price, the electricity price in the valley range is lower than 80% of the daily average electricity price, and the range between the two is classified as the flat peak range. Through this process, the periods of high and low electricity prices are identified, and the electricity price structure of different regions is adapted. The system operation mode is divided according to the results of the grid status determination and the electricity price interval division results, and multiple operation modes including charging priority mode, discharging priority mode, self-consumption priority mode, grid support mode, off-grid standby mode and emergency protection mode are determined. When the grid status is stable and the electricity price is in the low range, it automatically enters the charging priority mode. In this mode, the battery energy storage system uses low-priced electricity for charging to reduce subsequent electricity costs. When the grid status is still stable, but the electricity price enters the peak range, it switches to the discharge priority mode, so that the energy storage battery supplies power to the load to reduce the power purchase of the grid during the high electricity price period.If the photovoltaic power generation is high and can meet the load demand, it will enter the self-consumption priority mode, so that photovoltaic power generation will be supplied to the load first, and the battery will be charged when the excess power is allowed, so as to improve the self-generation and self-use rate of photovoltaic power generation. When the power grid is in a fluctuating state, it will enter the grid support mode, and the energy storage system will support the grid through the two-way power regulation function to alleviate the fluctuation problem of the power grid and enhance the stability of the system. If the power grid state deteriorates further and enters an abnormal state, it will automatically switch to the off-grid standby mode to ensure that the power supply of critical loads is not interrupted. In this mode, the energy storage system will rely on its own battery reserves to provide continuous power supply for important loads to avoid the failure of power terminals to operate normally due to power grid failures. When the system detects that the energy storage device itself is abnormal, such as excessive battery temperature, abnormal charge and discharge current, or communication failure, it will immediately enter the emergency protection mode and execute a series of safety protection measures, such as disconnecting the charge and discharge circuit, reducing the output power, or completely stopping the system operation to prevent equipment damage or safety accidents.

[0020] Step S200: Based on the grid state evaluation index and the system operation mode classification result, a multi-objective optimization algorithm is executed to obtain an energy scheduling plan including charging and discharging power instructions; Specifically, the three-phase hybrid inverter system is structurally simplified and divided into an interconnected system of the grid side, the DC bus side and the load side in order to establish a mathematical model. In this system, the grid side is connected to the DC bus through a bidirectional AC / DC converter, and the DC bus is connected to the energy storage unit and the photovoltaic input terminal at the same time, thus forming an energy interaction framework, while the load side is connected to the DC bus through a DC / AC converter to ensure that the load can obtain stable AC power. In order to quantify the operating state of the system, the system state vector is defined, which includes the DC bus voltage, battery state of charge, grid exchange power, load power and photovoltaic power generation power. These variables can fully describe the energy flow of the system and form a simplified model of the system. On this basis, the control input vector is defined, which includes the charging current command and the discharging current command. These two variables determine the energy inflow and outflow of the battery, and are therefore the core control quantities of the energy storage system scheduling. Combining the system state vector and the control input vector, the system control model is constructed, and based on this, the discrete time state equation of the system is established, thereby forming a system discrete time mathematical model suitable for digital control and predictive optimization. Since the energy management of the energy storage system involves three operating states: charging, discharging, and standby, the energy flow parameter is introduced into the mathematical model to enable it to reflect the dynamic characteristics of the system in different modes. The energy flow parameter is defined, which consists of the activation degree of the three states of charging, discharging, and standby. The activation degree of each state varies between 0 and 1 and satisfies the constraint relationship that its sum is 1 to ensure the uniqueness and identifiability of the system state. In order to accurately describe the operating characteristics of the energy storage system under different energy flow modes, a corresponding sub-model parameter matrix is ​​established for each energy flow state to obtain a discrete time system model. In the charging mode, the system parameter matrix needs to include the charging efficiency coefficient to ensure accurate modeling of the battery energy absorption process, while in the discharging mode, the system parameter matrix introduces the discharge efficiency coefficient to reflect the loss of the battery releasing energy to the load or the power grid. In the standby mode, considering the self-discharge characteristics of the battery, the corresponding parameter matrix should contain an equivalent coefficient that can describe the static loss of the battery, thereby ensuring the accurate description of the mathematical model for the standby state. Recursive least squares parameter update is performed on the discrete time system model to dynamically adjust the model parameters so that it can adapt to the influence of factors such as battery aging, grid fluctuations and environmental changes. The initial parameters are trained using historical data, and the model is gradually corrected based on real-time measurement data to ensure that the system model can maintain a high prediction accuracy under different working conditions. Since the change of DC bus voltage and the dynamic change of battery state of charge will affect the nonlinear characteristics of the charging and discharging process, these nonlinear factors are considered in the model update process, and the battery equivalent internal resistance model is introduced to correct the system's charging and discharging power calculation formula to ensure the accuracy of the optimization calculation. Through the model update process, the power flow parameters that accurately characterize the charging, discharging and standby states are obtained.According to the electric energy flow parameters, grid status evaluation indicators and system operation mode division results, a multi-objective optimization algorithm is executed to calculate the optimal energy scheduling plan including charging and discharging power instructions.

[0021] Based on the parameters of power flow and multi-objective optimization algorithm, the optimization objective function is defined. The objective function comprehensively considers the four core objectives of minimizing electricity cost, maximizing battery life, maximizing self-generation and self-consumption rate, and maximizing grid interaction support. At the same time, the weight coefficient in the optimization objective function is dynamically adjusted according to the system operation mode division results to ensure that the optimization objectives can be adaptively adjusted under different operation modes. According to the current grid state and operation mode, the dynamic weight coefficient is calculated in real time, and the weight adjustment strategy is optimized through historical data training. For example, in the charging priority mode, the weight coefficient of minimizing electricity cost increases, while in the discharging priority mode, the weight coefficient of battery life optimization increases, in the self-consumption priority mode, the weight coefficient of self-generation and self-consumption rate dominates, and in the grid support mode, the weight coefficient of grid interaction support will be greatly improved. The optimization algorithm reasonably allocates weights in the objective function according to the priority of different modes to ensure that the scheduling strategy meets the current operation requirements of the system. After determining the optimization objective function, the optimization constraints are set based on the dynamic weight coefficients to ensure that the optimization scheduling results can operate within physical and technical limitations. The optimization constraints mainly include battery state of charge restrictions, charging and discharging power restrictions, grid exchange power restrictions, and power balance constraints. Among them, the battery state of charge limit ensures that the battery's operating range does not exceed the minimum and maximum SOC values ​​to prevent overcharging or over-discharging; the charge and discharge power limit ensures that the battery's charge and discharge rate does not exceed the allowed safety range to protect the battery's health; the grid exchange power limit controls the amount of power exchange between the system and the grid to ensure that there is no unnecessary impact on the grid; the power balance constraint ensures the energy conservation between photovoltaic power generation, load power, battery charge and discharge power, and grid exchange power, thereby maintaining system stability. After the constraints are determined, the grid electricity price, load demand, and photovoltaic power generation are predicted based on the system parameter data set to obtain the system's future state prediction results. This process uses a time series forecasting method, in which the grid electricity price forecast is based on historical electricity price data and market trend modeling, and the electricity price fluctuation trend in the next 24 hours is predicted through the long short-term memory network (LSTM) or the autoregressive moving average model (ARIMA); the load demand forecast uses a hybrid forecasting method based on regression analysis and neural networks to obtain accurate load demand estimates by analyzing historical load curves, time characteristics and weather factors; photovoltaic power generation forecasting uses irradiance, temperature and historical power generation data for modeling, and predicts future power generation distribution through recursive neural networks (RNN) or support vector regression (SVR) algorithms. Through these forecasting models, future grid electricity price changes, user load demand and photovoltaic power generation capacity can be obtained in advance. Based on the system future state prediction results and optimization constraints, the optimization problem is converted into a quadratic programming form and solved by the interior point method to obtain preliminary optimization results.The core of the quadratic programming optimization method is to minimize the objective function while satisfying all constraints, so that the final solved charging and discharging strategy can achieve optimal energy management under different operating modes. In this process, based on the electricity price, load demand and photovoltaic power generation forecast data for the next 24 hours, the optimal charging and discharging power allocation at different time steps is calculated, and the optimization problem is constructed through the quadratic programming model, in which the objective function contains multiple optimization sub-objectives and is adjusted using dynamic weight coefficients. The interior point method is used for solving, which can efficiently solve large-scale optimization problems and ensure the global optimality of the optimization results while ensuring computational stability. Through this optimization process, the preliminary optimization results that meet the grid status, energy storage requirements and economic goals are calculated to ensure that the charging and discharging scheduling strategy can be reasonably adjusted under different working conditions. The preliminary optimization results are processed in the rolling time domain to generate the final energy scheduling plan containing the charging and discharging power instructions. The rolling time domain processing is to execute only the first time step result obtained by the optimization calculation within a certain optimization window each time, and recalculate the sliding window in the next optimization to achieve continuous dynamic optimization. A 6-hour optimization window is used, and the optimization update is performed every 15 minutes to ensure that the optimization scheduling can continuously adapt to changes in the real-time operating environment. At the same time, in order to reduce the computational complexity, each rolling optimization only adjusts the charge and discharge power of the most recent time step, while the control variables of the remaining time steps are referenced by the previous optimization results. The energy scheduling plan after rolling optimization contains the charge and discharge power instructions for the next 24 hours, and is dynamically adjusted every 15 minutes to ensure that the energy storage system can achieve optimal scheduling under complex power grid environments and fluctuating load demands.

[0022] Step S300: convert the energy scheduling plan into charge and discharge current control parameters, and input them into the control structure composed of the prediction control layer and the real-time response control layer in the three-phase hybrid inverter to obtain power output data; Specifically, the charge and discharge power instructions in the energy scheduling plan are converted to obtain the charge and discharge current control parameters. Since the relationship between the power and current of battery charging and discharging is affected by the DC bus voltage, the calculation of the charging current and the discharging current needs to consider the change of the instantaneous bus voltage to ensure the accuracy of the current calculation. The charging current is obtained by the ratio of the power instruction to the real-time bus voltage, and the discharging current also depends on the current bus voltage, and is corrected in combination with the battery equivalent internal resistance to generate accurate charge and discharge current control parameters. The charge and discharge current control parameters are input into the control structure of the three-phase hybrid inverter, which consists of a predictive control layer and a real-time response control layer to ensure that the control system can predict the system state in advance and respond quickly to changes in the power grid and load. Among them, the predictive control layer adopts a bidirectional long short-term memory network, which uses historical data to predict future charge and discharge current requirements, and obtains the system dynamic characteristics through time series modeling, thereby calculating a reasonable current reference value. The current reference value is passed to the real-time response control layer, which adopts a double closed-loop control structure, in which the inner loop is the current loop, which is responsible for the precise regulation of the current, and the outer loop is the power loop, which is used to maintain the system power balance and optimize the power output characteristics. Through the double-layer control method, accurate charging and discharging control is achieved in different operating modes, and it has strong dynamic adaptability. In the predictive control layer, the bidirectional long short-term memory network is trained by the historical data of the grid state, load demand, photovoltaic power generation and battery state of charge, and predicts and calculates in combination with the current system state to obtain the future current reference value. Since the bidirectional LSTM network considers the time series characteristics of the past and the future at the same time, it can accurately predict the current change trend in a complex energy storage system environment and effectively reduce the impact of external disturbances on current control. After the prediction calculation is completed, the current reference value is input into the real-time response control layer, in which the current loop is optimized offline using a proportional integral controller, and is adjusted online in combination with real-time data during system operation to compensate for the errors caused by grid fluctuations, load changes and environmental influences, and obtain a more stable current control signal. Based on the current control signal, power tracking control is performed through the power loop, in which the power loop compares the reference power with the actual power and dynamically adjusts the current command according to the power error to ensure that the system output power meets the requirements of the optimization scheduling plan. In order to improve the anti-disturbance ability of power tracking, a feedforward compensation mechanism is introduced. This mechanism adjusts the current command in advance according to the real-time monitored load power and photovoltaic power changes to compensate for the power fluctuations caused by changes in external conditions, thereby improving the dynamic response capability of the system. Under the control of the power flow parameters, a seamless switching mechanism is designed to ensure a smooth transition between different operating modes and avoid current shocks or power fluctuations caused by mode switching. In the implementation of the seamless switching mechanism, an exponential decay function is used as a smooth transition strategy for the power command to ensure the smoothness of the power adjustment process.The exponential decay function sets the rate of power adjustment according to the time constant, so that the system can gradually adjust to the new working state when the charging and discharging mode is switched, without generating sudden shocks. When the system switches from charging mode to discharging mode, or from discharging mode to standby mode, the exponential decay function will gradually reduce or increase the power command to ensure a smooth transition of power output during the system switching process. The time constant of the exponential decay function is adaptively adjusted according to the battery type. For example, the time constant of the lithium iron phosphate battery is set to 200ms, while the time constant of the ternary lithium battery is set to 150ms to match the charging and discharging characteristics of different types of batteries. In order to improve the grid adaptability of the system, the control signal of the smooth switching is combined with the virtual impedance reduction control strategy to optimize the regulation ability of the grid exchange power. The virtual impedance reduction control strategy simulates the equivalent impedance characteristics of the energy storage system, so that the system can better respond to the fluctuation of the grid voltage, and optimizes the system's support effect on the grid by dynamically adjusting the impedance parameters. At the same time, for the grid exchange power, a power smoothing algorithm based on sliding average is applied. While stabilizing the power output of the grid, the algorithm adaptively adjusts the smoothing window size according to the grid state. For example, when the power grid is in a stable state, the power smoothing window is set to 5 minutes, and when the power grid is in a fluctuating state, the smoothing window is reduced to 2 minutes to improve the system's response speed to power grid fluctuations. After calculation and optimization of all control links, the final calculated power control signal is converted into actual power output data through the IGBT power module. During this process, the IGBT module accurately adjusts the switching frequency and pulse width according to the system control instructions to ensure that the quality of the inverter output power meets the needs of the power grid and load. During the entire power conversion process, the system continuously monitors the real-time data of the output power and compares it with the reference power to ensure that the final output power meets the requirements of the scheduling plan, thereby achieving efficient and stable operation of the energy storage system.

[0023] Step S400: Upload the power output data of multiple three-phase hybrid inverters to the smart management cloud platform, execute multi-machine parallel coordinated control, and obtain a distributed energy network management solution.

[0024] Specifically, the system parameter data sets, power flow parameters and power output data of multiple three-phase hybrid inverters are collected respectively, and uploaded to the smart management cloud platform through a secure encrypted transmission protocol to ensure that the data will not be maliciously tampered with or leaked during the transmission process. The frequency of data upload is graded according to the importance of the parameters. The upload cycle of key status parameters is set to 10 seconds to ensure real-time performance, while general operating parameters are uploaded once every minute, and historical statistical data are stored at intervals of 15 minutes for subsequent analysis. In terms of data storage, a distributed database is used for management, and the time series data collected at high frequency is efficiently processed in combination with time series database technology to form structured storage data. Based on structured storage data, a remote monitoring and control architecture is established, which includes a data display layer, an analysis and decision-making layer, and a remote control layer to ensure that the cloud platform can provide complete visualization, intelligent analysis and remote control functions. In the data display layer, real-time monitoring functions are provided through the Web interface and mobile applications, allowing users to view the operating status of the energy storage system at any time, including important information such as system topology diagram, power flow diagram, energy storage status indication, and grid interaction status, and through dynamic data visualization technology, the interface updates the system operation status in real time. At the analysis and decision-making level, big data technology is used to analyze power consumption patterns, and combined with historical data, the cyclical change trend of user loads is identified, while future energy demand and photovoltaic power generation are predicted, and energy management strategies are optimized based on machine learning algorithms. At the remote control level, the system allows authorized users to send control instructions through the cloud platform, including adjusting battery charging and discharging strategies, setting power allocation ratios, switching operating modes, etc., and provides a remote access mechanism based on authority management to ensure the security and reliability of control authority. By building a remote monitoring and control architecture, centralized management of multiple inverters is achieved. With the support of the remote monitoring and control architecture, multiple three-phase hybrid inverters in parallel are divided, and one is set as the master inverter, which is responsible for global resource scheduling. The remaining inverters are slave inverters, which receive instructions sent by the master inverter and perform corresponding operations. The master inverter obtains global data through the cloud platform, and calculates a reasonable power sharing ratio based on the capacity, load and energy storage status of each inverter, thereby forming an efficient master-slave control structure. In this structure, the master inverter is responsible for global energy optimization scheduling, while monitoring the operating status of the slave inverters and adjusting the power allocation strategy when necessary to ensure the stability and efficiency of the entire system. The slave inverter performs the corresponding charging and discharging tasks according to the instructions of the master inverter, and feeds back its operating status to the master inverter so that the master inverter can adjust the control strategy in real time to achieve dynamic optimization. Based on the master-slave control structure, the master-slave droop control algorithm is adopted to ensure balanced power distribution in the multi-machine parallel system.The droop control algorithm adjusts the frequency and voltage output characteristics of the inverter so that each inverter can automatically distribute power according to its own capacity and load conditions without the need for complex centralized control. The system dynamically adjusts the droop characteristic curve according to the rated power of each inverter to ensure that the power can be reasonably distributed among multiple inverters when the load changes, while avoiding failure of a certain inverter due to overload. In actual applications, the master inverter continuously monitors the load changes of the entire system and adjusts the droop control parameters to optimize the power distribution effect. When the load changes suddenly, the system quickly adjusts the droop curve to ensure the stable operation of the entire parallel system, while reducing the impact of power fluctuations on the power grid, thereby improving the overall reliability of the system. Based on the multi-machine power balance distribution scheme, the virtual synchronous machine technology is introduced to enhance the grid adaptability of the parallel inverter cluster. By simulating the inertial characteristics of traditional synchronous generators, the virtual synchronous machine technology enables the energy storage system to provide a response capability similar to physical rotational inertia when connected to the grid, thereby improving the system's adaptability to grid frequency fluctuations. A synchronous inertia model is introduced into the controller of each inverter, and active frequency control is achieved through power regulation between inverters, so that the system can automatically adjust the output power when the grid is disturbed to alleviate the impact of grid fluctuations on the energy storage system. The system uses CAN bus for multi-machine communication to ensure that the information exchange between the master inverter and the slave inverter is sufficiently real-time and reliable. The communication cycle of the CAN bus is set to 10 milliseconds to ensure that all inverters can quickly share the operating status and control instructions and respond within milliseconds to achieve accurate multi-machine coordinated control. In the distributed energy network management solution, the operation safety in the case of communication interruption is considered. When a communication failure occurs on the CAN bus, the system automatically switches to independent operation mode, that is, each inverter operates independently according to local measurement data and preset strategies to ensure that the entire energy storage system can still maintain normal operation even in the case of communication failure. When the master inverter detects a communication anomaly, the local control mode will be triggered, and each slave inverter will continue to operate according to the latest power allocation instruction, and dynamically adjust the charging and discharging strategy according to the locally measured load conditions to minimize the impact of communication interruption on system operation. At the same time, the system attempts to re-establish the communication connection and re-enters the master-slave control mode after the communication is restored to ensure that the system can quickly return to normal operation.

[0025] In the embodiment of the present invention, a high-precision, multi-frequency parameter real-time acquisition system is used to realize comprehensive monitoring of the power grid state, energy storage state, load demand and photovoltaic power generation, providing a reliable data basis for system operation and making the energy flow identification more accurate. The power grid state evaluation mechanism based on power grid harmonic content analysis and weighted average composite function, combined with the electricity price interval division technology of fuzzy clustering algorithm, enables the system to intelligently perceive the change of power grid state and automatically adjust the operation mode according to the characteristics of electricity price. The parameterized representation method of power flow is introduced to express the system matrix and input matrix as functions of three states: charging, discharging and standby, which greatly simplifies the controller design and accurately describes the dynamic characteristics of the system under different power flows. A multi-objective optimization algorithm is used for energy scheduling, comprehensively considering multiple objectives such as electricity cost, battery life, self-generation and self-use rate and power grid interaction support, and dynamically adjusting the weight coefficient according to the system operation mode to achieve the optimal configuration of resources. A two-level control structure combining the predictive control layer and the real-time response control layer was designed. With the smooth transition strategy and power smoothing algorithm, the system oscillation and grid interaction power fluctuations were effectively suppressed, ensuring the stable operation of the three-phase hybrid inverter under various working conditions. The remote monitoring of system operation data and the coordinated control of multiple machines in parallel were realized through the intelligent management cloud platform. The master-slave droop control algorithm and virtual synchronous machine technology were adopted to enhance the stability and adaptability of the distributed energy network, especially the autonomous operation capability when communication was interrupted.

[0026] In a specific embodiment, the process of executing step S100 may specifically include the following steps: The voltage, frequency, phase and electricity price information of the three-phase power grid are sampled to obtain the power grid parameter measurement results; the state of charge, DC bus voltage, charge and discharge current and battery temperature of the battery pack are monitored to obtain the energy storage module parameter measurement results; the power consumption of the user's three-phase or single-phase load is collected to obtain the load demand monitoring results; the output voltage, current, power and environmental irradiance and temperature of the photovoltaic array are measured to obtain the photovoltaic power generation monitoring results; The grid parameter measurement results, energy storage module parameter measurement results, load demand monitoring results and photovoltaic power generation monitoring results are transmitted to the data processing unit, and preprocessed by denoising, outlier detection and missing value interpolation to obtain preprocessed data; Based on the preprocessed data, the power flow direction is identified by comparing the real-time values ​​of photovoltaic power generation, load demand power, battery charging and discharging power, and grid exchange power to obtain the system parameter data set; The power grid status is analyzed according to the system parameter data set to obtain the power grid status evaluation index and system operation mode division results.

[0027] Specifically, the sensors installed on the grid side capture three-phase voltage, frequency and phase information at a high sampling rate, and combine with smart meters to obtain electricity price information in real time. The update frequency of these data is usually in milliseconds to ensure that the system accurately reflects the changes in the state of the grid. At the same time, the current, voltage and temperature sensors inside the energy storage module continuously collect the state of charge of the battery, and calculate the battery health status through the built-in management system to predict the available capacity and performance degradation of the battery. The power consumption of the user's load is accurately monitored to grasp the real-time demand of the load. The load measurement module records the power consumption of three-phase or single-phase loads, identifies the peak power consumption period of the load, and forms the user's load change pattern. At the same time, in order to ensure the application effect of the system in the photovoltaic power generation system, the output voltage, current, power, environmental irradiance and temperature of the photovoltaic array are measured to obtain the real-time status of photovoltaic power generation. The photovoltaic monitoring module can record the working status of photovoltaic modules at a high frequency, and analyze the influence of environmental factors in combination with meteorological sensors. For example, on cloudy or rainy days, the output power of the photovoltaic array drops significantly, while on sunny days, it can generate more electricity. These data can help the system optimize the utilization strategy of photovoltaic power and improve the self-generation and self-use rate. The grid parameter measurement, energy storage module parameter measurement, load demand monitoring and photovoltaic power generation monitoring data are transmitted to the data processing unit, and a series of preprocessing is performed on the original data to improve the accuracy and reliability of the data. In this process, the denoising technology is used to smooth the data to eliminate abnormal fluctuations caused by sensor noise or environmental interference. In order to prevent the influence of outliers caused by measurement errors, threshold detection and statistical methods are used to detect outliers on the data, and interpolation technology is used to fill the missing data caused by signal loss. After the data is preprocessed, the power flow direction identification is performed based on these data to analyze the energy distribution of the current energy storage system. The core of power flow direction identification is to compare the real-time values ​​of photovoltaic power generation power, load demand power, battery charging and discharging power, and grid exchange power, and judge the current energy flow mode. The determination of the direction of electric energy flow can help the system optimize the energy storage management strategy and provide alarm information when necessary. For example, if the system detects that the battery charging current is too large and the photovoltaic power generation power is insufficient, it means that there is an abnormality in the system and it needs to be checked. After the electric energy flow direction is identified, the grid state is analyzed and the grid state evaluation index is calculated to determine the operating status of the grid. For example, by calculating the stability, phase consistency, frequency fluctuation rate and harmonic content of the three-phase grid voltage, a comprehensive score of the grid state is formed, which can reflect the health of the grid. In some scenarios, when the grid frequency fluctuates greatly or the harmonic content exceeds the set threshold, it is determined that the grid is in a fluctuating state and corresponding measures are taken, such as limiting the power of the grid to charge the battery, or reducing the rate at which the battery discharges to the grid, to reduce the impact on the grid.At the same time, combined with real-time electricity price information, the economic analysis of the power grid status is carried out. For example, when the power grid price is at a low point, even if the power grid status evaluation index is low, it is preferred to purchase electricity from the power grid for charging to reduce the overall operating cost. Based on these data, the system operation mode is divided. For example, when the power grid is in good condition and the electricity price is at a low point, the charging priority mode is entered, and when the power grid is in poor condition and the battery is fully charged, the discharge priority mode is entered, thereby achieving optimal energy storage management.

[0028] In a specific embodiment, the process of analyzing the power grid state according to the system parameter data set to obtain the power grid state evaluation index and the system operation mode classification result may specifically include the following steps: The three-phase voltage amplitude, phase difference, frequency and fluctuation rate of the power grid in the system parameter data set are continuously monitored to obtain the basic parameter monitoring results of the power grid, and the basic parameter monitoring results of the power grid are Fourier transformed to obtain the harmonic content analysis results of the power grid; Based on the analysis results of the harmonic content of the power grid, a composite function calculation of weighted average is performed to obtain the evaluation index of the power grid state; According to the grid state evaluation index, the grid state is divided into a stable state, a fluctuating state and an abnormal state, and a grid state determination result is obtained; Perform fuzzy clustering on the real-time electricity price data in the system parameter data set to obtain the electricity price interval division result; The system operation modes are divided according to the grid status judgment results and the electricity price range division results, and the system operation mode division results including charging priority mode, discharging priority mode, self-consumption priority mode, grid support mode, off-grid standby mode and emergency protection mode are obtained.

[0029] Specifically, high-precision sensors are used to collect key parameters on the power grid side in real time, and the data acquisition unit is used to continuously record them to ensure that the system can accurately reflect the operating status of the power grid. A high-sampling rate data acquisition module is used to capture the instantaneous changes of voltage, phase, and frequency at a millisecond sampling frequency, and the fluctuation rate is calculated using a short-term sliding window technology to measure the dynamic stability of the power grid. The monitoring results of the basic parameters of the power grid are Fourier transformed to extract the harmonic content of the power grid and analyze the impact of high-order harmonics on the stability of the power grid. The collected power grid signals are converted from the time domain to the frequency domain, and the amplitude distribution of each harmonic is calculated to identify the harmonic distortion of the power grid. The harmonic characteristics of the power grid are analyzed by Fourier transform to identify whether the power grid is disturbed by nonlinear loads, and corresponding optimization measures are taken. Based on the analysis results of the harmonic content of the power grid, a weighted average composite function calculation is performed to comprehensively measure the power grid state and determine the evaluation index of the power grid state. According to the three-phase voltage stability, phase consistency, frequency fluctuation rate, and harmonic distortion, the scores of each sub-indicator are calculated respectively, and these sub-indicators are combined into a unified power grid state evaluation value by the weighted average method. The operation state of the power grid is divided based on the grid state evaluation index to determine whether the grid is currently in a stable state, a fluctuating state, or an abnormal state. When the grid state evaluation index is in a high range, the grid is considered to be stable. At this time, the energy storage system performs charging and discharging operations according to the scheduling plan normally without affecting the safety of the grid. If the grid state evaluation index is in a medium range, it indicates that the grid has a certain degree of fluctuation, and the energy scheduling strategy needs to be adjusted to reduce the impact on the grid. For example, when the grid state is in a fluctuating state, the energy storage system reduces the rate of power injection into the grid to prevent the power change from causing further instability to the grid. When the grid state evaluation index drops to a lower range, indicating that the grid is in an abnormal state, the system takes emergency measures, such as switching to off-grid mode, to ensure stable power supply to important loads, while avoiding unstable operation of the energy storage system due to grid failure. After completing the grid state determination, the electricity price information is analyzed to optimize the economic operation strategy of the energy storage system. Fuzzy clustering is performed on the real-time electricity price data in the system parameter data set to divide the electricity price range and provide a basis for subsequent energy scheduling. Using historical electricity price data and current market electricity price information, the changing trend of electricity prices is identified, and the electricity price is divided into three intervals: peak, flat and valley through fuzzy clustering algorithm. According to the results of the grid state judgment and the results of the electricity price interval division, the operation mode of the energy storage system is determined to ensure that the system can not only meet the stability requirements of the grid operation, but also optimize the economic benefits of the energy storage system. The system operation modes include charging priority mode, discharging priority mode, self-consumption priority mode, grid support mode, off-grid standby mode and emergency protection mode. When the grid state is stable and the electricity price is in the valley range, the system enters the charging priority mode, giving priority to using low-priced electricity to charge the battery, so as to reduce the cost of purchasing electricity during future high electricity price periods.When the power grid is stable and the electricity price is in the peak range, the system enters the discharge priority mode, giving priority to releasing the energy of the energy storage battery to the load to reduce the amount of electricity purchased during the high electricity price period. When the photovoltaic power generation is sufficient and the power grid is stable, the system enters the self-consumption priority mode, so that the photovoltaic power is supplied to the load first, and the energy storage is charged to reduce the amount of electricity purchased by the power grid and improve the utilization rate of renewable energy. When the power grid is in a fluctuating state, the system switches to the grid support mode. In this mode, the energy storage system adjusts the charging and discharging power to reduce the impact of grid fluctuations on the load. In the abnormal state of the power grid, the system automatically enters the off-grid standby mode to ensure that the power supply to critical loads is not interrupted. When the system detects that there is an abnormality in the energy storage device itself, such as excessive battery temperature or abnormal charging and discharging current, the system enters the emergency protection mode and takes a series of safety measures, such as disconnecting the charging and discharging circuit or reducing power output, to prevent equipment damage or safety accidents.

[0030] In a specific embodiment, the process of executing step S200 may specifically include the following steps: The three-phase hybrid inverter system is simplified into an interconnected system of the grid side, the DC bus side and the load side, and the system state vector is defined to obtain a simplified system model; Based on the simplified system model, a control input vector including a charging current command and a discharging current command is defined to obtain a system control model; According to the system control model, the discrete time state equation of the system is established to obtain the discrete time mathematical model of the system, and the flow direction of electric energy is introduced into the discrete time mathematical model of the system to obtain a parameterized system model, where the flow direction of electric energy includes charging, discharging and standby; For each energy flow direction in the parameterized system model, a corresponding sub-model parameter matrix is ​​established to obtain a discrete time system model, which includes a charging efficiency coefficient in the charging state, a discharging efficiency coefficient in the discharging state, and reflects the self-discharge characteristics in the standby state; The recursive least squares method is used to update the parameters of the discrete-time system model, and the nonlinear relationship between the DC bus voltage change and the battery state of charge is considered to obtain the energy flow parameters that characterize the charging, discharging and standby states. According to the electric energy flow parameters, grid status evaluation indicators and system operation mode division results, a multi-objective optimization algorithm is executed to obtain an energy scheduling plan including charging and discharging power instructions.

[0031] Specifically, the three-phase hybrid inverter system is simplified to describe the energy flow and control logic of the system in a more intuitive way. The system is divided into three main parts: the grid side, the DC bus side, and the load side. The grid side is connected to the DC bus through a bidirectional AC / DC converter, the DC bus side is connected to the energy storage battery and the photovoltaic input terminal, and the load side is connected to the DC bus through a DC / AC converter, so that the entire system can flexibly switch the direction of power flow under different working conditions. Under this framework, the core variables of the system include the DC bus voltage, the battery state of charge, the grid exchange power, the load power, and the photovoltaic power generation power. These variables together constitute the system state vector, thereby establishing a simplified model of the system for subsequent control and optimization calculations. Based on the simplified model, the control input variables are defined to determine the regulation strategy of the system. Since the main control goal of the energy storage system is to manage the charging and discharging process, the control input vector includes the charging current command and the discharging current command. These two variables directly determine the charging and discharging rate of the battery and affect the change of the DC bus voltage. In order to describe the dynamic behavior of the system, the control model of the system is established based on the control input vector and the state variable, and the discrete time state equation of the system is constructed to obtain a complete mathematical description. In this process, considering the energy exchange relationship between the grid side, DC bus side and load side, the discrete time mathematical model of the system needs to comprehensively consider the battery charging and discharging characteristics, grid interaction power and load dynamic requirements, and incorporate these factors into the state update equation to ensure that the model can accurately describe the system characteristics under different operating modes. The energy flow direction parameter is introduced into the discrete time mathematical model of the system, so that the system can be dynamically adjusted according to different operating modes. The energy flow direction parameter includes three states: charging, discharging and standby. The activation degree of each state is represented by a normalized parameter, and the sum of these parameters is always 1 to ensure that the system is only in one of these three states at any time. For example, in the charging mode, the charging flow direction parameter is close to 1, while the discharge flow direction parameter and the standby flow direction parameter are close to 0; in the discharge mode, the discharge flow direction parameter is dominant, while the charging flow direction parameter and the standby flow direction parameter are small; in the standby state, all flow direction parameters are evenly distributed to reflect that the system is in a static equilibrium state. By introducing the energy flow direction parameter, the system state equation can flexibly adapt to different operating modes, thereby improving the adaptability of the control strategy. On this basis, corresponding sub-model parameter matrices are established for different states of power flow to construct a discrete-time system model. In the charging state, the system parameter matrix needs to include the charging efficiency coefficient to describe the conversion loss during the battery energy absorption process, while in the discharging state, the system parameter matrix needs to consider the discharge efficiency to accurately calculate the proportion of energy released by the battery to the load or the power grid. For the standby state, since the battery still has a certain self-discharge loss, the parameter matrix needs to include the self-discharge characteristics to ensure that the system can accurately estimate the actual available energy of the battery during long-term operation.In order to ensure the long-term accuracy of the discrete-time system model, its parameters are updated online. Therefore, the recursive least squares method is used to dynamically adjust the model parameters to adapt to environmental changes. In the calculation process of the recursive least squares method, the correlation coefficients of the charging, discharging and standby states in the parameter matrix are adjusted based on real-time data, so that the model can maintain a high prediction accuracy during long-term operation. Due to the nonlinear relationship between the DC bus voltage and the battery state of charge, the system needs to introduce nonlinear correction terms when updating parameters to ensure the accuracy of the optimization calculation. For example, under high load conditions, the nonlinear effect of the battery discharge rate causes large fluctuations in the bus voltage. Therefore, additional compensation terms need to be introduced during the parameter update process to correct the impact of changes in the battery equivalent internal resistance on the dynamic characteristics of the system. After obtaining accurate power flow parameters, combined with the grid state evaluation index and the operation mode division results, a multi-objective optimization algorithm is executed to calculate the optimal charging and discharging power instructions. The core of the optimization algorithm is to comprehensively consider multiple objectives, including minimizing electricity costs, maximizing battery life, maximizing self-generation and self-use rates, and maximizing grid interaction support, and dynamically adjust the weight coefficients so that the optimization objectives can adapt to different operation modes. In the optimization calculation process, based on the grid electricity price, load demand and photovoltaic power generation forecast data for the next 24 hours, the optimal charging and discharging power at different time steps is calculated, and the optimization problem is solved using the quadratic programming method. Through multi-objective optimization calculation, an energy dispatch plan containing charging and discharging power instructions for the next 24 hours is generated, and dynamic adjustments are made in combination with the rolling optimization strategy during actual operation to ensure that the energy storage system can achieve optimal energy management under different working conditions.

[0032] In a specific embodiment, the execution step executes a multi-objective optimization algorithm according to the electric energy flow direction parameter, the grid state evaluation index and the system operation mode division result, and the process of obtaining an energy scheduling plan including the charging and discharging power instructions may specifically include the following steps: Based on the electric energy flow parameters and the multi-objective optimization algorithm, the optimization objective function is defined, and the weight coefficient in the optimization objective function is dynamically adjusted according to the system operation mode division result to obtain the dynamic weight coefficient; Setting optimization constraints based on dynamic weight coefficients, including battery state of charge constraints, charging and discharging power constraints, grid exchange power constraints, and power balance constraints; Based on the system parameter data set, the grid electricity price, load demand and photovoltaic power generation are predicted to obtain the system future state prediction results; Based on the prediction results of the future state of the system and the optimization constraints, the optimization problem is transformed into a quadratic programming form and solved by the interior point method to obtain the preliminary optimization results; The preliminary optimization results are processed in the rolling time domain to generate an energy scheduling plan including charging and discharging power instructions.

[0033] Specifically, the core optimization objectives of the energy storage system are clarified, including minimizing electricity costs, maximizing battery life, maximizing the self-generation and self-use rate of photovoltaic power, and maximizing the support for grid interaction. The priorities of these objectives are different in different operating modes. During the optimization calculation process, the weight coefficients in the optimization objective function are dynamically adjusted according to the division results of the system operation mode, so as to ensure that the optimization strategy adapts to different working conditions. For example, in the charging priority mode, priority is given to using low electricity price periods for battery charging, thereby reducing the overall electricity cost, so the weight of minimizing electricity costs needs to be increased accordingly; while in the discharge priority mode, it is ensured that the battery supplies power to the load during high electricity price periods, thereby reducing the amount of electricity purchased by the grid, so as to increase the weight of the battery discharge strategy. In the self-consumption priority mode, the optimization goal focuses on the self-generation and self-use rate of photovoltaic power to minimize dependence on the grid, while in the grid support mode, the system needs to optimize the grid interaction support so that the energy storage system can provide effective frequency and voltage regulation capabilities when the grid fluctuates, so the weight of grid interaction support will be increased in this mode. Through the dynamic weight adjustment mechanism, the optimization algorithm can adapt to different system requirements and achieve the best energy management effect under various operating conditions. Optimization constraints are set based on dynamic weight coefficients to ensure that the calculated scheduling strategy meets both the safe operation requirements of physical equipment and the energy balance relationship. The optimization constraints include battery state of charge restrictions to ensure that the battery will not be overcharged and discharged, thereby protecting the long-term service life of the battery; charging and discharging power restrictions to ensure that the battery will not experience high charging and discharging rates in a short period of time, thereby avoiding overheating or overload damage; grid exchange power restrictions to prevent the system from injecting or extracting too much power from the grid to avoid affecting the stability of the grid; power balance constraints to ensure that the system can achieve energy conservation at any time, thereby ensuring that the load demand can be reasonably distributed by photovoltaic power generation, battery discharge and grid power supply. The future system state is predicted based on historical data to obtain future grid electricity prices, load demand and photovoltaic power generation. The grid electricity price forecast adopts the time series analysis method, which analyzes the electricity price data over the past period of time and combines the market trend model to predict the electricity price fluctuations within the next 24 hours. The load demand forecast is based on historical electricity consumption data, weather data and user behavior patterns, and uses machine learning algorithms for forecasting. Photovoltaic power generation prediction uses environmental sensor data, including solar irradiance, temperature, weather conditions and other factors, combined with the characteristic curve of photovoltaic modules, and uses neural networks or support vector regression algorithms to predict future photovoltaic power generation capacity. Through these prediction methods, the electricity price, load demand and photovoltaic power generation information for the next 24 hours are accurately obtained. Based on the prediction results of the future state of the system and the optimization constraints, the optimization problem is converted into a quadratic programming form and solved by the interior point method.The quadratic programming method is suitable for solving optimization problems with quadratic objective functions and linear constraints. It can ensure that the calculated scheduling results are optimal in the global scope and meet all constraints. In this process, the input of the optimization calculation includes the electricity price forecast, load demand forecast, photovoltaic power generation forecast, and the system's power flow parameters and optimization constraints for the next 24 hours. These data are used to construct the quadratic programming problem and the interior point method is used to solve it. The calculated optimization results provide reasonable charging and discharging power instructions for the energy storage system, so that it can operate according to the optimal strategy in the next 24 hours to maximize economic benefits and system stability. The preliminary optimization results are only the optimal solutions calculated based on static data. During the actual operation, the grid state, load demand and photovoltaic power generation will change dynamically. Therefore, the preliminary optimization results are processed in the rolling time domain to generate the final charging and discharging power instructions. In the rolling optimization process, a 6-hour optimization window is used, and the optimization update is performed every 15 minutes. That is, in each calculation, only the charging and discharging power instructions of the current time step are executed, and the optimization results of the subsequent time steps will be used as a reference for the next optimization to ensure that the system can adjust the scheduling strategy in real time. Through the rolling optimization method, the optimal scheduling effect can be maintained under uncertain operating conditions, and intelligent management of the energy storage system can be achieved.

[0034] In this embodiment, a multi-objective optimization algorithm is executed according to the electric energy flow direction parameters, the grid state evaluation index and the electricity price interval division results to obtain the process of energy scheduling plan including charging and discharging power instructions, which further includes the step of applying a two-layer deep reinforcement learning agent model, including: constructing a reinforcement learning environment model based on the system parameter data set, defining the battery state of charge, the grid state, the electricity price level, the load demand and the photovoltaic power generation power as the state space, setting the charging and discharging power as the action space, and constructing a reward function according to the electricity cost savings, the battery life impact, the grid stability and the spontaneous self-use rate to obtain the environmental interaction basis; designing a high-level strategy network for the environmental interaction basis, using a deep recurrent neural network architecture, inputting a 24-hour forecast data sequence, and outputting a long-term decision sequence with a time scale of hours to obtain a long-term planning strategy; designing a low-level execution network based on the long-term planning strategy, using a deep convolutional neural network architecture, inputting a sliding time window containing the real-time data of the last 30 minutes, and outputting a short-term control action with a time scale of minutes to obtain a short-term execution strategy; implementing a high-level strategy network and a low-level execution network based on The priority experience replay training mechanism gives samples with large timing difference errors a higher replay probability, and adopts the target network fixing technology to reduce training instability, and obtains the network optimization parameters; the hierarchical time abstraction method is applied to the network optimization parameters, so that the high-level network is triggered once every 4 hours and the low-level network is executed once every 5 minutes, and a hierarchical decision tree structure is constructed. The high-level and low-level network communications are connected through the option framework to obtain the timing coordination model; the multi-agent collaborative decision-making mechanism is introduced into the timing coordination model, each three-phase hybrid inverter is regarded as an independent agent, and the graph neural network based on the attention mechanism is used to establish the communication channel between agents, and the group collaborative optimization structure is obtained; based on the group collaborative optimization structure, the dynamic adaptation mechanism of the environment is developed, the deviation between the predicted value and the actual environmental change is compared, the accuracy weight of the power grid state assessment is dynamically adjusted, and the meta-learning method is used to quickly adapt to the new environment to obtain the adaptive learning model; the decision results output by the adaptive learning model are integrated with the quadratic planning optimization results, and a weighted fusion mechanism is adopted to generate more accurate charging and discharging power instructions, and an energy scheduling plan optimized by deep reinforcement learning is obtained.

[0035] In a specific embodiment, the process of executing step S300 may specifically include the following steps: Perform charge and discharge current conversion on the charge and discharge power instructions in the energy scheduling plan to obtain charge and discharge current control parameters; The charge and discharge current control parameters are input into the control structure of the three-phase hybrid inverter, which is composed of a prediction control layer and a real-time response control layer. The prediction control layer is a bidirectional long short-term memory network, and the real-time response control layer is a double closed-loop control structure, in which the inner loop is a current loop and the outer loop is a power loop. In the prediction control layer, the current prediction calculation of the bidirectional long short-term memory network is performed to obtain the current reference value, and the current reference value is input into the real-time response control layer. The current loop is optimized offline and adjusted online adaptively to obtain the current control signal; Based on the current control signal, power tracking is performed through the power loop, the actual power is compared with the reference power and the current command is dynamically adjusted. At the same time, a feedforward compensation mechanism is introduced according to the load and photovoltaic power changes to obtain an anti-disturbance power command; A seamless switching mechanism is designed based on the numerical value of the power flow direction parameter, and an exponential decay function is used as a smooth transition strategy corresponding to the anti-disturbance power command to obtain a smooth switching control signal; The control signal of smooth switching is combined with the virtual impedance drop control strategy, a power smoothing algorithm based on sliding average is applied to the grid exchange power, and the power output data is output through the IGBT power module.

[0036] Specifically, the current value required for charging and discharging is calculated based on the real-time DC bus voltage. Since the relationship between the charging and discharging power and current of the energy storage battery is affected by the bus voltage fluctuation, the current calculation method is dynamically adjusted to ensure accurate energy conversion. By real-time monitoring of the DC bus voltage and combining the internal equivalent resistance of the battery, the charging and discharging current control parameters that are more in line with the actual situation are calculated and input into the control structure of the three-phase hybrid inverter. The control structure consists of a predictive control layer and a real-time response control layer, in which the predictive control layer uses a bidirectional long short-term memory network for current prediction calculation, while the real-time response control layer uses a dual closed-loop control structure, including an inner current loop and an outer power loop, to ensure the accuracy and dynamic adaptability of power control. In the predictive control layer, the bidirectional long short-term memory network predicts the current reference value required at future times by analyzing the past and future grid states, load demand, and photovoltaic power generation power data. The current reference value is input into the real-time response control layer, where the current loop is optimized offline using a proportional-integral controller and adjusted online in combination with real-time data to compensate for deviations caused by grid fluctuations, load changes, and environmental impacts, thereby generating an accurate current control signal. In this way, the system flexibly adjusts the charging and discharging strategies in different operating modes to ensure the safety and energy conversion efficiency of the battery. After obtaining the current control signal, power tracking control is performed through the power loop to ensure that the actual power output is consistent with the reference power. The power loop compares the measured power with the reference power in real time and adjusts the current command based on the error, so that the system can accurately control the output power and reduce the power deviation caused by load changes or photovoltaic power generation fluctuations. In order to improve the system's anti-disturbance capability, a feedforward compensation mechanism is introduced into the power loop. This mechanism adjusts the current command in advance according to the changing trends of the load power and photovoltaic power generation to compensate for the power fluctuations that occur. For example, when the system detects that the photovoltaic power generation power increases, the feedforward compensation mechanism will appropriately reduce the power purchase from the grid or reduce the battery discharge power to optimize the energy utilization efficiency, thereby improving the system's dynamic adaptability. At the same time, in order to ensure that the system switches smoothly and reliably between charging, discharging and standby modes, a seamless switching mechanism is designed based on the numerical value of the power flow parameters, and an exponential decay function is used as a smooth transition strategy corresponding to the anti-disturbance power command to ensure the stability of current changes. In this mechanism, the time constant of the exponential decay function is dynamically adjusted according to the current power flow state to adapt to different switching requirements. For example, when the system switches from charging mode to discharging mode, the exponential decay function can gradually adjust the power command to prevent the battery or load from being unstable due to current mutations. The smooth transition strategy can effectively reduce power shocks and avoid system oscillations caused by mode switching, thereby improving the reliability of the overall system. The control signal of smooth switching is combined with the virtual impedance drop control strategy to optimize the regulation ability of the power grid exchange power.The virtual impedance drop control strategy simulates the equivalent impedance characteristics of the energy storage system, enabling the system to better respond to grid voltage fluctuations, and optimizes the system's support effect on the grid by dynamically adjusting the impedance parameters. At the same time, for the grid exchange power, a power smoothing algorithm based on sliding average is applied to reduce the impact of power fluctuations during the charging and discharging process on the grid. The power smoothing algorithm adaptively adjusts the smoothing window size according to the grid state. For example, when the grid is stable, the power smoothing window is set to a longer time interval to ensure the stability of the system operation. When the grid fluctuates, the system will shorten the smoothing window to improve the response speed to grid changes. After calculation and optimization of all control links, the final calculated power control signal is converted into actual power output data through the IGBT power module to ensure that the inverter can operate according to the optimal control strategy and maintain the best energy management effect in different operating modes.

[0037] In a specific embodiment, the process of executing step S400 may specifically include the following steps: The system parameter data sets, power flow parameters and power output data of multiple three-phase hybrid inverters are uploaded to the smart management cloud platform respectively, and the encrypted transmission protocol is used for distributed data storage to obtain structured storage data; Based on structured storage data, a data display layer is established, which includes a real-time monitoring function through a Web interface and a mobile application, an analysis and decision-making layer that uses big data technology to analyze power consumption patterns, and a remote control layer that allows authorized users to send control commands, thereby obtaining a remote monitoring and control architecture. A remote monitoring and control architecture is used to divide multiple three-phase hybrid inverters in parallel. One inverter is set as the master inverter responsible for global resource scheduling, and the rest are slave inverters that receive and execute instructions. The power sharing ratio is calculated according to the capacity, load and energy storage status of each three-phase hybrid inverter to obtain a master-slave control structure. Based on the master-slave control structure, the master-slave droop control algorithm is executed, and the droop characteristic curve is dynamically adjusted according to the inverter capacity to obtain a multi-machine balanced power distribution solution; Virtual synchronous machine technology is introduced into the multi-machine balanced power distribution scheme, and multi-machine communication is realized through the CAN bus. When the communication is interrupted, it automatically switches to the independent operation mode to obtain a distributed energy network management solution.

[0038] Specifically, the system parameter data set, power flow parameters, and power output data of each inverter are uploaded to the smart management cloud platform to achieve global coordinated control. In order to ensure the security and reliability of data transmission, an encrypted transmission protocol is adopted. After all parameters are signed and encrypted at the data acquisition end, they are distributedly transmitted through a secure communication protocol and stored as structured data in the cloud. This storage method uses a time series database combined with distributed file storage to ensure the efficiency of data query, while providing high fault tolerance, so that the system can maintain normal operation even in the event of partial server failure. Since these data include real-time power exchange information, grid status monitoring results, energy storage system charging and discharging status, and load demand changes, the storage structure needs to have high throughput and support historical data backtracking for subsequent energy optimization calculations and abnormal analysis. After completing data upload and storage, a remote monitoring and control architecture is established based on structured storage data to provide data visualization, intelligent analysis, and remote control functions. The remote monitoring and control architecture includes a data display layer, an analysis and decision layer, and a remote control layer. The data display layer provides real-time monitoring functions through a Web interface and mobile applications, allowing users to view the system's operating status at any location, including power flow, energy storage system status, grid interaction power, and load conditions, while providing historical data query and trend analysis functions. The analysis and decision layer uses big data technology to analyze the system's operating mode, identifies power consumption patterns through machine learning algorithms, and combines grid price fluctuations, photovoltaic power generation capacity, and load demand forecast results to provide the energy storage system with the best energy management strategy. The remote control layer allows authorized users to send control commands, including adjusting charging and discharging strategies, setting inverter power allocation ratios, switching operating modes, etc., and ensures the legitimacy and security of remote commands through a security mechanism based on authority management. The establishment of this architecture enables the system to perform remote monitoring and scheduling in a highly intelligent manner and achieve optimal energy storage management under different operating conditions. With the support of the remote monitoring and control architecture, multiple three-phase hybrid inverters in parallel are divided to establish a master-slave control structure. In order to achieve efficient energy scheduling, one inverter is set as the master inverter, which is responsible for the scheduling of global resources, while the remaining inverters are slave devices, receiving the instructions of the master inverter and performing corresponding operations. The master inverter calculates the power sharing ratio of all inverters based on the global data obtained by the cloud platform, and dynamically adjusts the power allocation strategy according to the capacity, current load and energy storage status of each inverter to ensure the stable operation of the system. For example, when the battery state of charge of one inverter is high and the battery power of another inverter is low, the master inverter assigns the inverter with higher power to undertake more discharge tasks to balance the overall charge state of the system. At the same time, the slave inverter outputs power according to the scheduling instructions of the master inverter and provides real-time feedback on its operating status so that the master inverter can adjust the scheduling strategy when necessary.The master-slave control structure ensures the coordination of the multi-machine parallel system and improves the energy management efficiency of the overall system. Based on the master-slave control structure, the master-slave droop control algorithm is adopted to achieve balanced power distribution of multiple inverters. The droop control algorithm adjusts the voltage-power and frequency-power characteristic curves of the inverter so that each inverter can dynamically adjust the power output according to its own capacity without centralized control. When the system load increases, the frequency decreases, and the inverter automatically increases the power output according to the droop control strategy. When the load decreases, the frequency increases and the inverter reduces the power output accordingly. Through this mechanism, the system realizes automatic load distribution without relying on centralized scheduling and improves the stability of the overall system. In order to optimize the multi-machine balanced power distribution scheme, the virtual synchronous machine technology is introduced to enhance the inverter's adaptability to grid fluctuations. The virtual synchronous machine technology simulates the inertia and damping characteristics of the synchronous generator in the inverter control strategy, so that the inverter can provide a response capability similar to the physical rotation inertia when the grid frequency fluctuates, thereby improving the dynamic stability of the system. For example, when the grid experiences an instantaneous load change, the inverter temporarily stores or releases energy to reduce the fluctuation amplitude of the grid frequency and enhance the anti-interference ability of the system. At the same time, in order to ensure that the information exchange between the master inverter and the slave inverter is sufficiently real-time and reliable, the CAN bus is used for multi-machine communication, and the communication cycle is set to 10 milliseconds to ensure that all inverters can quickly share the operating status and control instructions and respond within milliseconds. In actual operation, the system needs to deal with the problem of communication interruption. When the CAN bus fails, the system automatically switches to the independent operation mode to obtain a distributed energy network management solution. In the independent operation mode, each inverter operates independently according to local measurement data and preset strategies to ensure that the energy storage system can still maintain normal operation in the event of communication failure. For example, when the master inverter detects a communication anomaly, the local control mode is triggered, so that each slave inverter continues to operate according to the latest power allocation instruction, and dynamically adjusts the charging and discharging strategy according to its own load conditions to reduce the impact of communication interruption on system operation. At the same time, the system attempts to re-establish the communication connection and re-enters the master-slave control mode after the communication is restored to ensure that the system can quickly return to normal operation.

[0039] See also Figure 2 , Figure 2 A schematic block diagram of the structure of an intelligent energy storage control device 200 based on bidirectional energy management provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the intelligent energy storage control device 200 based on bidirectional energy management includes: The real-time acquisition module 210 is used to collect grid parameters, energy storage module parameters, load demand and photovoltaic power generation data in real time and analyze the grid status to obtain grid status evaluation indicators and system operation mode classification results; The multi-objective optimization module 220 is used to execute a multi-objective optimization algorithm based on the grid state evaluation index and the system operation mode classification result to obtain an energy scheduling plan including charging and discharging power instructions; The prediction control module 230 is used to convert the energy scheduling plan into charge and discharge current control parameters, and input them into the control structure composed of the prediction control layer and the real-time response control layer in the three-phase hybrid inverter to obtain power output data; The parallel coordination control module 240 is used to upload the power output data of multiple three-phase hybrid inverters to the smart management cloud platform, perform multi-machine parallel coordination control, and obtain a distributed energy network management solution.

[0040] Through the cooperation of the above components, a high-precision, multi-frequency parameter real-time acquisition system is used to realize comprehensive monitoring of the grid state, energy storage state, load demand and photovoltaic power generation, providing a reliable data basis for system operation and making the energy flow identification more accurate. The grid state evaluation mechanism based on grid harmonic content analysis and weighted average composite function, combined with the electricity price interval division technology of fuzzy clustering algorithm, enables the system to intelligently perceive the change of grid state and automatically adjust the operation mode according to the characteristics of electricity price. The parameterized representation method of power flow is introduced to express the system matrix and input matrix as functions of three states: charging, discharging and standby, which greatly simplifies the controller design and accurately describes the dynamic characteristics of the system under different power flow. A multi-objective optimization algorithm is used for energy scheduling, comprehensively considering multiple objectives such as electricity cost, battery life, self-generation and self-use rate and grid interaction support, and dynamically adjusting the weight coefficient according to the system operation mode to achieve the optimal allocation of resources. A two-level control structure combining the predictive control layer and the real-time response control layer was designed. With the smooth transition strategy and power smoothing algorithm, the system oscillation and grid interaction power fluctuations were effectively suppressed, ensuring the stable operation of the three-phase hybrid inverter under various working conditions. The remote monitoring of system operation data and the coordinated control of multiple machines in parallel were realized through the intelligent management cloud platform. The master-slave droop control algorithm and virtual synchronous machine technology were adopted to enhance the stability and adaptability of the distributed energy network, especially the autonomous operation capability when communication was interrupted.

[0041] See also Figure 3 , Figure 3 A schematic block diagram of the structure of an intelligent energy storage control device 300 based on bidirectional energy management provided in an embodiment of the present application, wherein the intelligent energy storage control device 300 based on bidirectional energy management includes a processor 301 and a memory 302, wherein the processor 301 and the memory 302 are connected via a device bus 303, wherein the memory 302 may include a non-volatile storage medium and an internal memory.

[0042] The non-volatile storage medium can store a computer program. The computer program includes program instructions, and when the program instructions are executed by the processor 301, the processor 301 can execute any of the above-mentioned intelligent energy storage control methods based on bidirectional energy management.

[0043] The processor 301 is used to provide computing and control capabilities to support the operation of the entire intelligent energy storage control device 300 based on bidirectional energy management.

[0044] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can execute any of the above-mentioned intelligent energy storage control methods based on bidirectional energy management.

[0045] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the intelligent energy storage control device 300 based on bidirectional energy management involved in the scheme of the present application. The specific intelligent energy storage control device 300 based on bidirectional energy management may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0046] It should be understood that the processor 301 may be a central processing unit (CPU), and the processor 301 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0047] It should be noted that technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working process of the intelligent energy storage control device 300 based on bidirectional energy management described above can refer to the corresponding process of the aforementioned intelligent energy storage control method based on bidirectional energy management, and will not be repeated here.

[0048] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0049] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0050] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent energy storage control method based on bidirectional energy management, characterized in that: include: Real-time collection and grid status analysis of grid parameters, energy storage module parameters, load demand and photovoltaic power generation data are carried out to obtain grid status evaluation indicators and system operation mode classification results; Based on the grid state evaluation index and the system operation mode division result, a multi-objective optimization algorithm is executed to obtain an energy scheduling plan including charging and discharging power instructions; The energy scheduling plan is converted into charge and discharge current control parameters, and input into a control structure composed of a prediction control layer and a real-time response control layer in a three-phase hybrid inverter to obtain power output data; The power output data of multiple three-phase hybrid inverters are uploaded to the intelligent management cloud platform, and multi-machine parallel coordinated control is performed to obtain a distributed energy network management solution.

2. The intelligent energy storage control method based on bidirectional energy management according to claim 1 is characterized in that: The real-time collection of grid parameters, energy storage module parameters, load demand and photovoltaic power generation data and grid status analysis are performed to obtain grid status evaluation indicators and system operation mode classification results, including: The voltage, frequency, phase and electricity price information of the three-phase power grid are sampled to obtain the power grid parameter measurement results; the state of charge, DC bus voltage, charge and discharge current and battery temperature of the battery pack are monitored to obtain the energy storage module parameter measurement results; the power consumption of the user's three-phase or single-phase load is collected to obtain the load demand monitoring results; the output voltage, current, power and environmental irradiance and temperature of the photovoltaic array are measured to obtain the photovoltaic power generation monitoring results; The grid parameter measurement results, the energy storage module parameter measurement results, the load demand monitoring results and the photovoltaic power generation monitoring results are transmitted to a data processing unit, and preprocessed by denoising, outlier detection and missing value interpolation to obtain preprocessed data; Based on the preprocessed data, the power flow direction is identified by comparing the real-time values ​​of photovoltaic power generation power, load demand power, battery charging and discharging power, and grid exchange power to obtain a system parameter data set; The power grid state is analyzed according to the system parameter data set to obtain power grid state evaluation indicators and system operation mode classification results.

3. The intelligent energy storage control method based on bidirectional energy management according to claim 2 is characterized in that: The analyzing the power grid state according to the system parameter data set to obtain the power grid state evaluation index and the system operation mode classification result includes: Continuously monitoring the three-phase voltage amplitude, phase difference, frequency and fluctuation rate of the power grid in the system parameter data set to obtain the basic parameter monitoring results of the power grid, and performing Fourier transform on the basic parameter monitoring results of the power grid to obtain the harmonic content analysis results of the power grid; Performing a weighted average composite function calculation based on the power grid harmonic content analysis result to obtain a power grid state evaluation index; According to the power grid state evaluation index, the power grid state is divided into a stable state, a fluctuating state and an abnormal state, and a power grid state determination result is obtained; Performing fuzzy clustering on the real-time electricity price data in the system parameter data set to obtain an electricity price interval division result; The system operation mode is divided according to the grid state determination result and the electricity price range division result, and the system operation mode division results including charging priority mode, discharging priority mode, self-consumption priority mode, grid support mode, off-grid standby mode and emergency protection mode are obtained.

4. The intelligent energy storage control method based on bidirectional energy management according to claim 3 is characterized in that: The multi-objective optimization algorithm is executed based on the grid state evaluation index and the system operation mode division result to obtain an energy scheduling plan including charging and discharging power instructions, including: The three-phase hybrid inverter system is simplified into an interconnected system of the grid side, the DC bus side and the load side, and the system state vector is defined to obtain a simplified system model; Based on the simplified system model, a control input vector including a charging current instruction and a discharging current instruction is defined to obtain a system control model; Establishing a system discrete time state equation according to the system control model to obtain a system discrete time mathematical model, and introducing the electric energy flow direction into the system discrete time mathematical model to obtain a parameterized system model, wherein the electric energy flow direction includes charging, discharging and standby; For each electric energy flow direction in the parameterized system model, a corresponding sub-model parameter matrix is ​​established to obtain a discrete time system model, wherein the charging state includes a charging efficiency coefficient, the discharging state includes a discharging efficiency coefficient, and the standby state reflects the self-discharge characteristics; Executing a recursive least squares method to update the parameters of the discrete-time system model, and considering the nonlinear relationship between the DC bus voltage change and the battery state of charge, to obtain power flow parameters characterizing the charging, discharging and standby states; According to the electric energy flow direction parameters, the grid status evaluation index and the system operation mode division results, a multi-objective optimization algorithm is executed to obtain an energy scheduling plan including charging and discharging power instructions.

5. The intelligent energy storage control method based on bidirectional energy management according to claim 4 is characterized in that: The multi-objective optimization algorithm is executed according to the electric energy flow direction parameter, the grid state evaluation index and the system operation mode division result to obtain an energy scheduling plan including charging and discharging power instructions, including: Defining an optimization objective function based on the electric energy flow direction parameter and the multi-objective optimization algorithm, and dynamically adjusting a weight coefficient in the optimization objective function according to the system operation mode division result to obtain a dynamic weight coefficient; Setting optimization constraints based on the dynamic weight coefficients, the optimization constraints including battery state of charge restriction, charge and discharge power restriction, grid exchange power restriction and power balance restriction; Based on the system parameter data set, grid electricity price, load demand and photovoltaic power generation are predicted to obtain a prediction result of the future state of the system; Based on the prediction result of the future state of the system and the optimization constraints, the optimization problem is converted into a quadratic programming form and solved by an interior point method to obtain a preliminary optimization result; The preliminary optimization results are subjected to rolling time domain processing to generate an energy dispatch plan including charging and discharging power instructions.

6. The intelligent energy storage control method based on bidirectional energy management according to claim 5 is characterized in that: The energy scheduling plan is converted into a charge and discharge current control parameter and input into a control structure composed of a prediction control layer and a real-time response control layer in a three-phase hybrid inverter to obtain power output data, including: Performing charge and discharge current conversion on the charge and discharge power instructions in the energy scheduling plan to obtain charge and discharge current control parameters; Inputting the charge and discharge current control parameters into a control structure composed of a prediction control layer and a real-time response control layer in a three-phase hybrid inverter, wherein the prediction control layer is a bidirectional long short-term memory network, and the real-time response control layer is a double closed-loop control structure, wherein the inner loop is a current loop and the outer loop is a power loop; In the prediction control layer, a current prediction calculation of a bidirectional long short-term memory network is performed to obtain a current reference value, and the current reference value is input into the real-time response control layer, and an offline optimization and online adaptive adjustment are performed through the current loop to obtain a current control signal; Based on the current control signal, power tracking is performed through the power loop, actual power is compared with reference power and current instructions are dynamically adjusted, and a feedforward compensation mechanism is introduced according to load and photovoltaic power changes to obtain an anti-disturbance power instruction; A seamless switching mechanism is designed based on the numerical value of the electric energy flow direction parameter, and an exponential decay function is used as a smooth transition strategy corresponding to the anti-disturbance power instruction to obtain a control signal for smooth switching; The control signal of the smooth switching is combined with the virtual impedance drop control strategy, a power smoothing algorithm based on sliding average is applied to the grid exchange power, and power output data is output through the IGBT power module.

7. The intelligent energy storage control method based on bidirectional energy management according to claim 6 is characterized in that: The power output data of multiple three-phase hybrid inverters are uploaded to the smart management cloud platform, and multi-machine parallel coordinated control is performed to obtain a distributed energy network management solution, including: The system parameter data sets, power flow parameters and power output data of multiple three-phase hybrid inverters are uploaded to the smart management cloud platform respectively, and the encrypted transmission protocol is used for distributed data storage to obtain structured storage data; Based on the structured storage data, a data display layer is established, which includes providing real-time monitoring functions through a Web interface and a mobile application, an analysis and decision-making layer that uses big data technology to analyze power consumption patterns, and a remote control layer that allows authorized users to send control instructions, thereby obtaining a remote monitoring control architecture; The remote monitoring control architecture is adopted to divide multiple three-phase hybrid inverters in parallel, and one is set as the master inverter responsible for global resource scheduling, and the rest are slave inverters that receive and execute instructions, and the power sharing ratio is calculated according to the capacity, load condition and energy storage state of each three-phase hybrid inverter, so as to obtain a master-slave control structure; Based on the master-slave control structure, a master-slave droop control algorithm is executed, and the droop characteristic curve is dynamically adjusted according to the inverter capacity to obtain a multi-machine balanced power distribution scheme; The virtual synchronous machine technology is introduced into the multi-machine balanced power distribution scheme, and multi-machine communication is realized through the CAN bus. When the communication is interrupted, it automatically switches to the independent operation mode to obtain a distributed energy network management solution.

8. An intelligent energy storage control device based on bidirectional energy management, characterized in that: Used to execute the intelligent energy storage control method based on bidirectional energy management as described in any one of claims 1 to 7, the intelligent energy storage control device based on bidirectional energy management comprises: Real-time acquisition module, used for real-time acquisition of grid parameters, energy storage module parameters, load demand and photovoltaic power generation data and grid status analysis, to obtain grid status evaluation indicators and system operation mode classification results; A multi-objective optimization module, used to execute a multi-objective optimization algorithm based on the grid state evaluation index and the system operation mode division result to obtain an energy scheduling plan including charging and discharging power instructions; A prediction control module, used for converting the energy scheduling plan into charge and discharge current control parameters, and inputting the parameters into a control structure composed of a prediction control layer and a real-time response control layer in a three-phase hybrid inverter to obtain power output data; The parallel coordination control module is used to upload the power output data of multiple three-phase hybrid inverters to the intelligent management cloud platform, perform multi-machine parallel coordination control, and obtain a distributed energy network management solution.

9. An intelligent energy storage control device based on bidirectional energy management, characterized in that: The intelligent energy storage control device based on bidirectional energy management comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instruction in the memory so that the intelligent energy storage control device based on bidirectional energy management executes the intelligent energy storage control method based on bidirectional energy management as described in any one of claims 1-7.

Images