Electric energy distribution method, device and equipment based on energy storage equipment

Abstract

The present invention relates to the technical field of electric energy distribution, and discloses an electric energy distribution method, device and equipment based on energy storage equipment, which are used to improve the accuracy of electric energy distribution of energy storage equipment and improve the stability of household electricity consumption. The method includes: initializing energy management of hybrid energy storage equipment to obtain an initial electric energy distribution strategy combination; performing electric signal data feature analysis on the hybrid energy storage equipment to obtain a first electric signal feature set and a second electric signal feature set; obtaining historical energy consumption data and environmental variables and performing energy demand forecasting to obtain target energy demand forecasting data; inputting the first electric signal feature set, the second electric signal feature set and the target energy demand forecasting data into a Markov decision model to perform charge and discharge priority allocation and electric energy distribution strategy adjustment, and generate a target electric energy distribution strategy combination.

Description

Technical Field

[0001] The present invention relates to the technical field of electric energy distribution, and in particular to an electric energy distribution method, device and equipment based on energy storage equipment. Background Art

[0002] Existing hybrid energy storage devices usually adopt static or preset power distribution strategies, which often cannot respond to changes in the external environment and the dynamic changes in the actual energy consumption of households in real time. For example, during the peak period of photovoltaic power generation, excess electricity is not effectively stored or utilized, resulting in energy waste; and during peak demand periods, the system may not be able to provide enough electricity, affecting the stability of household electricity consumption. Summary of the invention

[0003] The present invention provides an energy storage device-based electric energy distribution method, device and equipment, which are used to improve the accuracy of electric energy distribution of the energy storage device and improve the stability of household electricity consumption.

[0004] In a first aspect, the present invention provides an electric energy distribution method based on an energy storage device, the electric energy distribution method based on the energy storage device comprising:

[0005] Initializing energy management of a hybrid energy storage device to obtain an initial electric energy distribution strategy combination, wherein the hybrid energy storage device includes: a first energy storage device and a second energy storage device;

[0006] Performing an electric signal data feature analysis on the hybrid energy storage device based on the initial electric energy allocation strategy combination to obtain a first electric signal feature set of the first energy storage device and a second electric signal feature set of the second energy storage device;

[0007] Acquire historical energy consumption data and environmental variables, and perform energy demand forecasting on the hybrid energy storage device to obtain target energy demand forecast data;

[0008] The first electric signal feature set, the second electric signal feature set and the target energy demand forecast data are input into a preset Markov decision model to perform charge and discharge priority allocation and electric energy distribution strategy adjustment to generate a target electric energy distribution strategy combination.

[0009] In a second aspect, the present invention provides an electric energy distribution device based on an energy storage device, the electric energy distribution device based on the energy storage device comprising:

[0010] An initialization module is used to initialize energy management of a hybrid energy storage device to obtain an initial electric energy distribution strategy combination, wherein the hybrid energy storage device includes: a first energy storage device and a second energy storage device;

[0011] An analysis module, configured to perform an electric signal data feature analysis on the hybrid energy storage device based on the initial electric energy allocation strategy combination to obtain a first electric signal feature set of the first energy storage device and a second electric signal feature set of the second energy storage device;

[0012] A prediction module, used to obtain historical energy consumption data and environmental variables, and to predict energy demand for the hybrid energy storage device to obtain target energy demand prediction data;

[0013] A generation module is used to input the first electrical signal feature set, the second electrical signal feature set and the target energy demand forecast data into a preset Markov decision model to perform charge and discharge priority allocation and electric energy distribution strategy adjustment to generate a target electric energy distribution strategy combination.

[0014] The third aspect of the present invention provides an electric energy distribution device based on an energy storage device, 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 electric energy distribution device based on the energy storage device executes the above-mentioned electric energy distribution method based on the energy storage device.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for distributing electric energy based on an energy storage device.

[0016] In the technical solution provided by the present invention, by integrating historical energy consumption data and environmental variables to predict energy demand, it is possible to reflect changes in energy demand in real time and adaptively adjust the energy allocation strategy. This combination of real-time and adaptability enables the system to always maintain optimal energy efficiency and cost-effectiveness under different environmental conditions and changes in consumer demand. The Markov decision model is used to dynamically allocate charging and discharging priorities and adjust the power allocation strategy. Through the combination of target power allocation strategies, the relationship between energy supply and actual household needs can be more accurately matched, energy waste can be effectively reduced, and the service life of energy storage equipment can be optimized. Through the characteristic analysis of electrical signal data and intelligent strategy adjustment, the method can improve the response speed and accuracy of the system to external changes, thereby enhancing the overall stability of the system. Whether it is environmental conditions such as weather changes or fluctuations in the internal load of the home, the system can respond quickly to ensure the stability of the power supply. It can effectively utilize renewable energy such as solar energy and wind energy to reduce dependence on traditional power grids. Through accurate electrical signal feature analysis and target energy demand prediction, this method can more effectively utilize each unit of energy. Optimizing the charging and discharging process and time ensures sufficient energy when needed while avoiding overcharging or discharging, thereby improving overall energy efficiency, thereby improving the accuracy of power distribution based on energy storage devices and improving the stability of household electricity consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] 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.

[0018] Figure 1 A schematic diagram of an embodiment of a method for distributing electric energy based on an energy storage device in an embodiment of the present invention;

[0019] Figure 2 A schematic diagram of an embodiment of an electric energy distribution device based on an energy storage device in an embodiment of the present invention;

[0020] Figure 3 It is a schematic diagram of an embodiment of an electric energy distribution device based on an energy storage device in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] Embodiments of the present invention provide a method, apparatus and device for distributing electric energy based on energy storage devices. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0022] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the electric energy distribution method based on the energy storage device in the embodiment of the present invention includes:

[0023] Step S101, initializing energy management of a hybrid energy storage device to obtain an initial electric energy distribution strategy combination, the hybrid energy storage device comprising: a first energy storage device and a second energy storage device;

[0024] It is understandable that the execution subject of the present invention may be an energy distribution device based on an energy storage device, or may be a terminal or a server, which is not specifically limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0025] Specifically, the basic parameter space of the hybrid energy storage device is defined, which covers multiple charging and discharging parameters such as charging and discharging current, charging and discharging voltage, and charging and discharging time period, and each parameter has a corresponding value range. These parameters are the key factors controlling the charging and discharging behavior of the energy storage device, and their value range determines the performance and safety margin of the device in actual operation. The sine and cosine algorithm is used to generate uniformly distributed random values ​​for each charging and discharging parameter in the basic parameter space. The sine and cosine algorithm ensures that the generated random values ​​are not only random but also uniform through the characteristics of periodic functions, thereby covering all possible areas of the parameter space, exploring the parameter space to the greatest extent, and finding possible optimal or more effective charging and discharging strategies. According to the random value of each parameter, the first initial power distribution strategy of the first energy storage device in the hybrid energy storage device and the second initial power distribution strategy of the second energy storage device are generated. These two strategies are formulated for different types of energy storage devices respectively. Considering the characteristics of the device such as maximum withstand voltage, optimal working current, etc., the formulation of the strategy aims to use randomly generated parameter values ​​to explore the performance of each device under different working conditions to ensure maximum energy efficiency while ensuring device safety. According to the preset parameter search space and number of iterations, the initial power distribution strategy is analyzed for strategy fusion. Through iterative search and analysis, various possible strategy combinations are analyzed to evaluate their performance in practical applications, such as energy utilization, response speed, and system stability. Through iterative fusion analysis, a set of initial power distribution strategy combinations is obtained that not only considers the efficiency of individual devices but also optimizes the overall system performance.

[0026] Step S102: analyzing the electric signal data characteristics of the hybrid energy storage device based on the initial electric energy allocation strategy combination to obtain a first electric signal characteristic set of the first energy storage device and a second electric signal characteristic set of the second energy storage device;

[0027] Specifically, based on the initial power distribution strategy combination, the hybrid energy storage device is tested for energy supply and data is collected to obtain the first electrical signal data of the first energy storage device and the second electrical signal data of the second energy storage device. By actually running these devices and regulating them according to the initial power distribution strategy, electrical signal data reflecting the state and performance of each device are collected. Based on the first and second electrical signal data, a corresponding state matrix is ​​constructed to obtain a first state matrix and a second state matrix. The state matrix is ​​constructed by arranging the electrical signal data according to the time series and electrical signal characteristics (such as voltage, current, frequency, etc.) to form a mathematical model that can represent the change of the electrical signal state over time. The matrix not only reflects the state of the electrical signal at different time points, but also facilitates mathematical analysis and processing. The eigenvalues ​​of the first state matrix and the second state matrix are calculated to obtain the main factors driving the change of the electrical signal, and the main eigenvalues ​​of the matrix are extracted to determine which factors play a key role in the change of the electrical signal. The first characteristic difference and the second characteristic difference obtained are indicators of the most significant changes in the electrical signal data of each device, and these differences indicate the main change trends and patterns of the electrical signals. Based on the obtained characteristic differences, the corresponding electrical signal data is signal reorganized and signal features are extracted. Signal recombination refers to adjusting the combination of original electrical signal data according to feature differences to more clearly display key electrical signal changes; while signal feature extraction refers to extracting key information from the reorganized electrical signals, such as peak value, average value, fluctuation, etc., thereby obtaining the first electrical signal feature set of the first energy storage device and the second electrical signal feature set of the second energy storage device, recording the electrical signal performance characteristics of each device under specific operating and environmental conditions.

[0028] In this embodiment, the QR algorithm is used to calculate the eigenvalues ​​of the first and second state matrices. The QR algorithm is an efficient numerical method suitable for calculating the eigenvalues ​​of large-scale matrices. Through iterative decomposition and reconstruction processes, the eigenvalues ​​of the matrix are gradually approached. The first state matrix is ​​input into the QR algorithm, and the algorithm obtains multiple eigenvalues ​​of the first matrix through a series of orthogonalization and triangular decomposition steps. Similarly, the second state matrix is ​​processed by the same QR algorithm to obtain multiple eigenvalues ​​of the second matrix. The maximum value among these eigenvalues ​​is identified, and the maximum eigenvalue usually represents the main change direction of the matrix or the main dynamic characteristics of the system. Through mathematical analysis methods, such as traversal or optimization algorithms, the first maximum eigenvalue is identified from the eigenvalue set of the first state matrix, and the second maximum eigenvalue is identified from the eigenvalue set of the second state matrix. The geometric mean of each set of eigenvalues ​​is calculated. The geometric mean provides a method that takes into account the common influence of all eigenvalues, which can better reflect the overall distribution characteristics of the matrix eigenvalues ​​than the simple arithmetic mean. The geometric mean of the eigenvalues ​​of the first matrix and the geometric mean of the eigenvalues ​​of the second matrix are calculated by taking the logarithm, averaging and then exponentially obtaining them to avoid numerical instability caused by excessive multiplication. The first maximum eigenvalue and the first characteristic geometric mean are calculated to obtain the first characteristic difference, and the second maximum eigenvalue and the second characteristic geometric mean are calculated to obtain the second characteristic difference. The characteristic difference reflects the non-uniformity or potential instability of the system state, which helps to adjust and optimize the power distribution strategy. Through the characteristic difference, the operation strategy of the energy storage device can be adjusted more accurately to optimize the energy distribution and utilization efficiency of the entire system.

[0029] Step S103: Acquire historical energy consumption data and environmental variables, and perform energy demand forecasting on the hybrid energy storage device to obtain target energy demand forecasting data;

[0030] Specifically, the energy consumption of hybrid energy storage equipment is analyzed to obtain historical energy consumption data, which includes key indicators such as energy usage at various time points in history and peak demand time. At the same time, environmental variables are monitored in real time, which may include temperature, humidity, sunshine conditions, etc. The historical energy consumption data is vector-coded and mapped, and the energy consumption data is converted into mathematically processable coding vectors through mathematical algorithms, so that these data can be effectively processed by machine learning models. The energy consumption coding vector is vector-fused with the environmental variable data. During the fusion process, the environmental variable data is also similarly coded to ensure the consistency of the data format and the effectiveness of data fusion. The target fusion coding vector is input into the preset two-layer long short-term memory network (LSTM) model. The two-layer LSTM model consists of forward and backward LSTM networks. This bidirectional structure can effectively capture the front-to-backward dependencies of time series data and enhance the model's understanding of time series data. In the first layer of the model, the forward LSTM network processes the target fusion encoding vector, extracts the time series dependency from the past to the future, and generates a forward hidden feature vector; at the same time, the backward LSTM network processes data from the future to the past, extracts the dependency on the other side of the time series, and generates a backward hidden feature vector. After that, the forward hidden feature vector and the backward hidden feature vector are vector-concatenated to obtain a concatenated hidden feature vector. The concatenated hidden feature vector is processed by the fully connected layer in the two-layer LSTM model, and the future energy demand is predicted using the learning ability of the neural network, and the target energy demand prediction data is output. The prediction results help the smart home energy storage system to more accurately adjust the energy allocation strategy to adapt to the predicted energy demand, thereby improving energy efficiency and the economy of the system.

[0031] Step S104: input the first electric signal feature set, the second electric signal feature set and the target energy demand forecast data into a preset Markov decision model to perform charge and discharge priority allocation and electric energy allocation strategy adjustment to generate a target electric energy allocation strategy combination.

[0032] Specifically, a first agent of the first energy storage device and a second agent of the second energy storage device are created, and the three basic components of the Markov decision model are defined for each agent: state, action, and reward function. The state defines the various situations that the system may be in, and the action is the behavior that the system can choose in a specific state. The reward function evaluates the results obtained after taking a certain action to guide the agent to make the best decision. On this basis, the electrical signal feature set of the first energy storage device and the target energy demand forecast data are integrated to construct a two-dimensional feature matrix. This matrix integrates the operating characteristics of the device and the external energy demand status, providing data support for the formulation of the power distribution strategy. The same method is also used for the second energy storage device, integrating its electrical signal feature set with the energy demand forecast data to construct a second two-dimensional feature matrix. The first two-dimensional feature matrix is ​​input into the Markov decision model, and the first agent is responsible for analyzing these data, and the first target power distribution strategy is obtained through model calculation. The strategy specifies the actions and priorities that the first energy storage device should choose under different states, aiming to maximize the overall benefits of the system, such as energy efficiency improvement and cost savings. At the same time, the second agent performs the same processing on the second two-dimensional feature matrix to obtain the second target power allocation strategy for the second energy storage device. The first agent and the second agent are assigned charging and discharging priorities. The energy supply and demand status of the two devices in a specific time period is evaluated to determine which device should be charged or discharged first, as well as the operation priority of each device in different time periods. The key to the correct setting of charging and discharging priorities is to ensure that resources can be effectively allocated when energy supply is insufficient or demand peaks, and to ensure the stable operation and efficiency of the home energy system. According to the charging and discharging priority information, the first and second target power allocation strategies are adjusted globally. This includes further refinement and synchronization of the two strategies to ensure that the strategies of the two energy storage devices can work together to jointly respond to changes in the energy demand of the family. Through the integration and optimization of strategies, a comprehensive target power allocation strategy combination is finally formed, which not only reflects the operation optimization of a single device, but also reflects the ability of the entire system to adaptively adjust in a changing environment, thereby achieving the optimal configuration of energy and maximizing the efficiency of use.

[0033] The first agent is imported into the Agent program, which is loaded with the operation logic and algorithm of the Markov decision model. The first two-dimensional feature matrix is ​​input into the Markov decision model. The feature data is analyzed by the model to determine the first candidate power distribution strategy learned by the first agent. Similarly, the second agent processes the second two-dimensional feature matrix through the same process to determine the second candidate power distribution strategy suitable for the second energy storage device. Based on the candidate strategy, each agent determines the action it should take at present. The first agent selects its first current action according to the first candidate power distribution strategy, and then sends this action to a virtual environment for execution. The virtual environment simulates the energy use and storage conditions in the real world so as to update the state of the agent, and the update reflects the new situation after the action is executed. The second current action of the second agent is determined according to the second candidate power distribution strategy, and the second current action is sent to the virtual environment for state update to obtain the second updated state. The reward feedback parameters of each agent are calculated based on the updated state and the preset reward function. The reward parameters are the key factors for evaluating the success of each action. They are based on multiple indicators, such as energy utilization efficiency, cost savings and system sustainability. The reward feedback parameters of the first agent are determined by its first updated state and the reward function, while the reward feedback parameters of the second agent are determined by the second updated state and the same reward function. The agents adjust their policies to optimize future action selection. The policy-action update gradient is calculated, which indicates how the policy should be adjusted to maximize future rewards. The first agent calculates the first policy-action update gradient based on its first reward feedback parameter, while the second agent performs the same calculation based on the second reward feedback parameter. Finally, based on the policy-action update gradient, each agent determines its final target power distribution strategy. The target strategy takes into account the operating characteristics and historical performance of each device, and integrates predictions of future energy demand and adaptability to environmental changes.

[0034] In the embodiment of the present invention, by integrating historical energy consumption data and environmental variables to predict energy demand, it is possible to reflect changes in energy demand in real time and adaptively adjust the energy allocation strategy. This combination of real-time and adaptability enables the system to always maintain optimal energy efficiency and cost-effectiveness under different environmental conditions and changes in consumer demand. The Markov decision model is used to dynamically allocate charging and discharging priorities and adjust the power allocation strategy. Through the combination of target power allocation strategies, the relationship between energy supply and actual household demand can be more accurately matched, energy waste can be effectively reduced, and the service life of energy storage equipment can be optimized. Through the characteristic analysis of electrical signal data and intelligent strategy adjustment, the method can improve the response speed and accuracy of the system to external changes, thereby enhancing the overall stability of the system. Whether it is environmental conditions such as weather changes or fluctuations in the internal load of the home, the system can respond quickly to ensure the stability of the power supply. It can effectively utilize renewable energy such as solar energy and wind energy to reduce dependence on traditional power grids. Through accurate electrical signal characteristic analysis and target energy demand prediction, this method can more effectively utilize each unit of energy. Optimizing the charging and discharging process and time ensures sufficient energy when needed while avoiding overcharging or discharging, thereby improving overall energy efficiency, thereby improving the accuracy of power distribution based on energy storage devices and improving the stability of household electricity consumption.

[0035] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0036] (1) defining a basic parameter space of a hybrid energy storage device, wherein the basic parameter space includes a plurality of charging and discharging parameters and corresponding value ranges, and the plurality of charging and discharging parameters include: charging and discharging current, charging and discharging voltage, and charging and discharging time period;

[0037] (2) Generate uniformly distributed random values ​​for each charge and discharge parameter in the basic parameter space using the sine and cosine algorithm;

[0038] (3) generating a first initial power distribution strategy for a first energy storage device and a second initial power distribution strategy for a second energy storage device in the hybrid energy storage device according to the random value of each charging and discharging parameter;

[0039] (4) According to the preset parameter search space and number of iterations, a strategy fusion analysis is performed on the first initial power allocation strategy and the second initial power allocation strategy to obtain an initial power allocation strategy combination.

[0040] Specifically, the basic parameter space of the hybrid energy storage device is defined. The basic parameter space includes multiple key charging and discharging parameters and their corresponding value ranges, including charging and discharging current, charging and discharging voltage, and charging and discharging time period. Each parameter has specific physical and technical limitations. For example, the range of charging and discharging current may be set between 0 and 100 amperes, the voltage range may be 100 to 400 volts, and the charging and discharging time period may be any period between 0 and 24 hours. Uniformly distributed random values ​​for each charging and discharging parameter are generated in the basic parameter space by the sine and cosine algorithm. The sine and cosine algorithm is a technique for generating uniformly distributed random numbers, which ensures that the generated values ​​are not only random but also uniformly cover the entire parameter range through the characteristics of periodic functions. This method is suitable for optimization problems that require exploring possibilities from a wide parameter space. Using these randomly generated parameter values, the first and second initial power distribution strategies are generated for the first energy storage device and the second energy storage device in the hybrid energy storage device, respectively. These strategies are set according to the characteristics of each device and randomly generated parameter values. For example, the first energy storage device may use a higher charging current and a shorter charging time to quickly respond to high-demand periods, while the second energy storage device may use a lower charging current and a longer charging time to provide a smoother energy output. Perform strategy fusion analysis. The strategies of the two energy storage devices are integrated and optimized to ensure that they can work together to maximize the energy efficiency and responsiveness of the entire system. By setting a preset parameter search space and number of iterations, different strategy combinations are continuously adjusted and tested to find the best power distribution strategy. The iterative process is similar to the training process in machine learning, which learns how to better configure and manage the charging and discharging behaviors of the two devices through continuous trial and error. For example, if it is found in an iteration that the charging strategy of the first energy storage device during high-demand periods causes excessive load on the grid, the system may adjust the strategy to reduce the charging current or adjust the charging time period. Similarly, if it is found that the second energy storage device is still charging at a high current during low-demand periods, the system may also adjust its strategy to more efficiently utilize available energy. Through iteration and adjustment, the final initial combination of power distribution strategies will be able to effectively support the energy needs of the home while taking into account cost-effectiveness and sustainable operation of the system. This process not only optimizes the individual performance of each device, but also enhances the synergy of the entire system through the integration of strategies, achieving optimal energy distribution of hybrid energy storage devices.

[0041] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0042] (1) performing energy supply testing and data collection on the hybrid energy storage device based on the initial electric energy distribution strategy combination to obtain first electrical signal data of the first energy storage device and second electrical signal data of the second energy storage device;

[0043] (2) constructing a first state matrix of the first energy storage device based on the first electrical signal data, and constructing a second state matrix of the second energy storage device based on the second electrical signal data;

[0044] (3) Calculating the eigenvalues ​​of the first state matrix to obtain a first eigendifference value, and calculating the eigenvalues ​​of the second state matrix to obtain a second eigendifference value;

[0045] (4) performing signal reorganization and signal feature extraction on the first electrical signal data according to the first characteristic difference value to obtain a first electrical signal feature set of the first energy storage device, and performing signal reorganization and signal feature extraction on the second electrical signal data according to the second characteristic difference value to obtain a second electrical signal feature set of the second energy storage device.

[0046] Specifically, the energy supply test and data collection of the hybrid energy storage device are performed based on the initial power distribution strategy combination. The electrical signal data of the first energy storage device and the second energy storage device are monitored and recorded when the system is running. For example, assuming that the first energy storage device is a high-capacity lithium-ion battery and the second energy storage device is a supercapacitor, these devices will generate different electrical signals during charging and discharging according to their respective operating characteristics and environmental conditions. In the actual test, sensors and data acquisition modules are used to record these electrical signal data, such as current intensity, voltage level, and charging and discharging time period. In this way, detailed information about the operating state of each device under a specific strategy is obtained. The electrical signal data is converted into a format that can be mathematically analyzed, and a state matrix of the first and second energy storage devices is constructed. The state matrix is ​​constructed by arranging the electrical signal data in a time series, each row represents a time point, and each column represents a different electrical signal feature, such as current and voltage. For example, if the current and voltage data of the first energy storage device during a charging cycle are recorded, these data can be organized into a matrix, in which each row corresponds to a measurement time point, and the columns represent current and voltage respectively. Perform eigenvalue calculation. Eigenvalue calculation is an important mathematical method used to analyze the main characteristics of matrices. By calculating the eigenvalues ​​of the first and second state matrices, numerical values ​​reflecting the dynamic characteristics of each system are obtained. These eigenvalues ​​help identify the most significant behavioral patterns in the operation of the device, such as under which conditions the voltage or current shows the largest fluctuations. Signal reorganization and feature extraction are performed based on the calculation results of the eigenvalues. Based on the calculated eigenvalues, the original electrical signal data is reorganized to more clearly reflect important patterns or trends. For example, if the eigenvalue analysis shows that the voltage fluctuation is the largest at a certain charging current, this information can be used to adjust the representation of the electrical signal data to make this fluctuation more prominent. Through reorganization, more specific signal features, such as peak voltage, average current, etc., are extracted to form a set of electrical signal features for the first and second energy storage devices. These feature sets provide a detailed overview of the electrical signal features for each device, which can be directly used for further development and optimization of power distribution strategies. For example, if it is found that the voltage stability of the first energy storage device is reduced when charging at high current, this discovery can prompt the system to develop a more sophisticated charging strategy to avoid operating under high load conditions, thereby extending battery life and improving system efficiency. Likewise, if data from the second energy storage device shows that it is able to maintain a higher voltage level during rapid discharge, this characteristic could be used to quickly release energy during peak demand moments, increasing the system's responsiveness.

[0047] In a specific embodiment, the execution step calculates the eigenvalue of the first state matrix to obtain the first eigendifference value, and calculates the eigenvalue of the second state matrix to obtain the second eigendifference value. The process may specifically include the following steps:

[0048] (1) performing eigenvalue calculation on the first state matrix based on the QR algorithm to obtain a plurality of eigenvalues ​​of the first matrix, and performing eigenvalue calculation on the second state matrix based on the QR algorithm to obtain a plurality of eigenvalues ​​of the second matrix;

[0049] (2) performing maximum identification on multiple first matrix eigenvalues ​​to obtain a first maximum eigenvalue, and performing maximum identification on multiple second matrix eigenvalues ​​to obtain a second maximum eigenvalue;

[0050] (3) performing geometric mean calculation on multiple eigenvalues ​​of the first matrix to obtain a first characteristic geometric mean value, and performing geometric mean calculation on multiple eigenvalues ​​of the second matrix to obtain a second characteristic geometric mean value;

[0051] (4) Calculate the difference between the first maximum eigenvalue and the first characteristic geometric mean to obtain a first characteristic difference, and calculate the difference between the second maximum eigenvalue and the second characteristic geometric mean to obtain a second characteristic difference.

[0052] Specifically, the QR algorithm is applied to the first state matrix and the second state matrix to calculate the eigenvalues. The QR algorithm is an effective numerical method for calculating the eigenvalues ​​of large matrices. The matrix is ​​decomposed into an orthogonal matrix (Q) and an upper triangular matrix (R) through an iterative process, and then these matrices are recombined to gradually approximate the eigenvalues ​​of the original matrix. For example, assume that the system has a state matrix representing the electrical signal of the first energy storage device, each row of the matrix represents the electrical signal at a different time point, and the column may include parameters such as current and voltage. After applying the QR algorithm, a set of eigenvalues ​​describing the characteristics of the electrical signal of the device is obtained. The largest eigenvalue is identified from the eigenvalues ​​of the first matrix, and the largest eigenvalue is similarly identified from the eigenvalues ​​of the second state matrix. These largest eigenvalues ​​represent the most dominant factors in the system dynamics, that is, the aspects of the system behavior that change the most. For example, if the largest eigenvalue comes from the data of the current change, this may indicate that the current change has the most significant impact on the system state. The geometric mean is calculated for all eigenvalues. The geometric mean is a statistical measure that is suitable for comparing values ​​of different scales. It is obtained by taking the logarithm of all eigenvalues, averaging them, and then exponentially. The difference between the maximum eigenvalue and the geometric mean is calculated. This difference (characteristic difference) provides an indicator of the dispersion of system characteristics. The larger the difference, the more obvious the dominance of certain behaviors or states of the system in terms of characteristics, which may require special attention or adjustment. For example, if the maximum eigenvalue of the first energy storage device is very different from the geometric mean, this may indicate that the device is very sensitive under certain operating conditions and its charging and discharging parameters need to be adjusted to prevent over-response or instability.

[0053] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0054] (1) Analyze the energy consumption of the hybrid energy storage device to obtain historical energy consumption data, and monitor the environmental variables of the hybrid energy storage device to obtain environmental variables;

[0055] (2) Perform vector coding mapping on historical energy consumption data to obtain energy consumption coding vector;

[0056] (3) Perform vector fusion on the energy consumption coding vector based on environmental variables to obtain the target fused coding vector;

[0057] (4) Input the target fusion encoding vector into the preset two-layer LSTM model;

[0058] (5) The forward LSTM network in the two-layer LSTM model is used to extract hidden features of the target fusion coding vector to obtain a forward hidden feature vector;

[0059] (6) The backward LSTM network in the two-layer LSTM model extracts hidden features from the forward hidden feature vector to obtain a backward hidden feature vector;

[0060] (7) Concatenating the forward hidden feature vector and the backward hidden feature vector to obtain a concatenated hidden feature vector;

[0061] (8) The energy demand is predicted by concatenating the hidden feature vector through the fully connected layer in the two-layer LSTM model to obtain the target energy demand prediction data.

[0062] Specifically, historical energy consumption data is collected from the system, including information such as power consumption, usage time, peak load period, etc., and environmental variables that affect energy consumption, such as temperature, humidity, light intensity, etc., are monitored at the same time. The collected data is converted into a form acceptable to the model through appropriate processing methods. For example, the historical energy consumption data and environmental variables are vector-encoded and mapped, and the actual measurement values ​​are converted into a vector form that can express the characteristics of the data. Vector encoding includes direct conversion of numerical values, and also includes standardization or normalization of these numerical values ​​to make them suitable for mathematical operations and model processing. For example, the power consumption is normalized according to the maximum and minimum consumption values, and the ambient temperature can be converted to a value within a standardized range. Based on the environmental variables, the energy consumption encoding vector is vector-fused, and the environmental variable encoding vector and the energy consumption encoding vector are merged into a single, more complex data structure to obtain the target fused encoding vector. All possible factors that affect energy consumption are integrated so that the model can learn and predict energy demand more comprehensively. The fused encoding vector is input into a preset two-layer long short-term memory (LSTM) model. The LSTM model is suitable for processing time series data and can learn long-term dependencies in the data. In the two-layer LSTM model, the first layer is a forward LSTM network that processes the target fusion encoding vector, extracts features from the beginning to the end of the time series, and captures the dependency from the past to the future. The second layer is a backward LSTM network that processes data from the end to the beginning of the time series, capturing the dependency from the future to the past. The bidirectional processing method enables the model to understand data from two time directions and enhances the ability to extract features. The extracted forward and backward hidden feature vectors are spliced ​​together to form a spliced ​​hidden feature vector that fully reflects the characteristics of the time series data. The splicing process ensures that the model can not only learn information from one direction of time, but also supplement the understanding of the time series from the other direction, making the prediction more accurate. Finally, the spliced ​​hidden feature vector is processed by the fully connected layer of the two-layer LSTM model to perform energy demand forecasting and obtain the target energy demand forecast data. For example, the model may predict that in the next few hours, based on the current environmental conditions and historical energy usage patterns, the household's electricity consumption will increase, so that the system can adjust the operation strategy of the energy storage device in advance to optimize energy use and cost.

[0063] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0064] (1) creating a first agent of the first energy storage device and a second agent of the second energy storage device, and defining the state, action, and reward function of the Markov decision model respectively;

[0065] (2) constructing a two-dimensional matrix for the first electrical signal feature set and the target energy demand forecast data to obtain a first two-dimensional feature matrix, and constructing a two-dimensional matrix for the second electrical signal feature set and the target energy demand forecast data to obtain a second two-dimensional feature matrix;

[0066] (3) importing the first intelligent agent and inputting the first two-dimensional feature matrix into the Markov decision model to analyze the power distribution strategy, thereby obtaining the first target power distribution strategy; at the same time, importing the second intelligent agent and inputting the second two-dimensional feature matrix into the Markov decision model to analyze the power distribution strategy, thereby obtaining the second target power distribution strategy;

[0067] (4) Allocate charging and discharging priorities for the first intelligent agent and the second intelligent agent to obtain charging and discharging priority information, and perform global energy allocation adjustments on the first target power allocation strategy and the second target power allocation strategy according to the charging and discharging priority information to generate a target power allocation strategy combination.

[0068] Specifically, a first agent of a first energy storage device and a second agent of a second energy storage device are created. The role of these agents is to perform independent decision-making processes for each device. After creating these agents, the components of an applicable Markov decision model are defined for them, including defining the state, action, and reward function of each agent. The state may include the current energy level, time, historical consumption pattern, and external environmental factors. The action may involve operations such as charging, discharging, or standby, and the reward function is based on the cost-effectiveness and energy efficiency of the action taken, aiming to guide the agent to select the most optimized power distribution strategy. The first electrical signal feature set and target energy demand forecast data collected by each device are processed to construct a two-dimensional feature matrix suitable for the Markov decision model. For example, the electrical signal feature set of the first energy storage device may include data such as voltage, current, and charging cycle, while the target energy demand forecast data provides possible energy demand patterns in the future. These data are integrated to construct a first two-dimensional feature matrix. Similarly, a second two-dimensional feature matrix is ​​constructed based on the relevant data of the second energy storage device. Each two-dimensional feature matrix is ​​input into the corresponding agent, which is analyzed by the Markov decision model to formulate or optimize the power distribution strategy. The first agent analyzes the first two-dimensional feature matrix to generate a first target power distribution strategy for the first energy storage device; at the same time, the second agent processes the second two-dimensional feature matrix to formulate a second target power distribution strategy for the second energy storage device. The first agent and the second agent are assigned charging and discharging priorities. The performance and requirements of the two energy storage devices under different operating conditions are evaluated and compared to determine which device should be charged or discharged first. For example, if the first energy storage device is mainly used to cope with energy demand during peak hours, while the second device is more used for daily low-load operation, the agent may decide to give priority to charging the second device during low-demand periods. Based on the charging and discharging priority information, the first and second target power distribution strategies are adjusted for global energy distribution to form a comprehensive target power distribution strategy combination. The strategy combination reflects the specific needs and operating priorities of each device and takes into account the energy efficiency and sustainable operation requirements of the overall system. For example, by adjusting the charging time and power distribution, the operating efficiency of the entire system can be maximized while meeting the energy requirements of all devices, ensuring that the system operates under the premise of efficiency and economy.

[0069] In a specific embodiment, the execution step of importing the first agent and inputting the first two-dimensional feature matrix into the Markov decision model to perform power distribution strategy analysis to obtain the first target power distribution strategy, and at the same time, importing the second agent and inputting the second two-dimensional feature matrix into the Markov decision model to perform power distribution strategy analysis to obtain the second target power distribution strategy may specifically include the following steps:

[0070] (1) Importing the first agent into the Agent program and inputting the first two-dimensional feature matrix into the Markov decision model to determine the first candidate power allocation strategy learned by the first agent; importing the second agent into the Agent program and inputting the second two-dimensional feature matrix into the Markov decision model to determine the second candidate power allocation strategy learned by the second agent;

[0071] (2) determining a first current action of the first agent according to the first candidate power allocation strategy, and sending the first current action to the virtual environment for state update to obtain a first updated state; determining a second current action of the second agent according to the second candidate power allocation strategy, and sending the second current action to the virtual environment for state update to obtain a second updated state;

[0072] (3) calculating a first reward feedback parameter of the first agent according to the first updated state and the reward function, and calculating a second reward feedback parameter of the second agent according to the second updated state and the reward function;

[0073] (4) calculating the first strategy action update gradient of the first agent according to the first reward feedback parameter, and calculating the second strategy action update gradient of the second agent according to the second reward feedback parameter;

[0074] (5) Determine a first target power allocation strategy of the first agent according to the first strategy action update gradient, and determine a second target power allocation strategy of the second agent according to the second strategy action update gradient.

[0075] Specifically, a corresponding agent is created for each energy storage device in the system. The agents are operated through the Agent program, and each agent has a corresponding decision model. The electrical signal data and energy demand forecast data of the first energy storage device and the second energy storage device are converted into two-dimensional feature matrices, which are provided as input data to the corresponding agents. For example, the first agent manages a lithium battery-based energy storage system, while the second agent manages a supercapacitor-based energy storage system, each of which has its own unique operating characteristics and energy demand pattern. Each agent learns from its corresponding two-dimensional feature matrix and determines a candidate power distribution strategy through a Markov decision model. In this process, the agent considers how to optimize the timing and magnitude of charging and discharging to cope with future changes in energy demand. The Markov decision model supports the agent in making decisions by defining states (such as current energy level, time and predicted energy demand), actions (such as charging, discharging or maintaining the current state) and rewards (based on cost, efficiency and reliability). After the agent determines the candidate strategies, these strategies are executed in a simulated environment. The virtual environment simulates actual operating conditions, allowing the agent to test the effectiveness of its strategy by implementing its current actions and observing the results. The execution of these actions and the feedback of the results will result in state updates, which provide the agent with new data points to further learn and adjust the strategy. After each state update, the agent calculates reward feedback parameters based on a predefined reward function. These parameters are key to evaluating the effectiveness of each action and reflect the impact of each operation on system efficiency, cost, and reliability. The agent uses this feedback to adjust its strategy to ensure the achievement of long-term goals, such as minimizing costs, maximizing energy utilization, or other business-related goals. By analyzing the reward feedback, the agent can calculate the policy action update gradient, which is a mathematical expression used to guide how to adjust the strategy to improve the expected reward of future actions. These update gradients help the agent find a better policy path in its decision space. Finally, based on the update gradients, the agent can determine its target power distribution strategy. These strategies are optimized operation plans designed to achieve the most efficient energy use and management for each energy storage device. In this way, each device can not only meet its specific energy needs, but also play the best role in the entire home energy system to ensure the continuity and efficiency of energy supply.

[0076] The above describes the electric energy distribution method based on the energy storage device in the embodiment of the present invention. The following describes the electric energy distribution device based on the energy storage device in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, an electric energy distribution device based on an energy storage device includes:

[0077] Initialization module 201, used to initialize energy management of hybrid energy storage device to obtain initial electric energy distribution strategy combination, the hybrid energy storage device includes: first energy storage device and second energy storage device;

[0078] An analysis module 202 is used to perform an electric signal data feature analysis on the hybrid energy storage device based on the initial electric energy allocation strategy combination to obtain a first electric signal feature set of the first energy storage device and a second electric signal feature set of the second energy storage device;

[0079] The prediction module 203 is used to obtain historical energy consumption data and environmental variables, and to predict the energy demand of the hybrid energy storage device to obtain target energy demand prediction data;

[0080] The generation module 204 is used to input the first electrical signal feature set, the second electrical signal feature set and the target energy demand forecast data into a preset Markov decision model to perform charge and discharge priority allocation and electric energy allocation strategy adjustment to generate a target electric energy allocation strategy combination.

[0081] Through the synergy of the above components, by integrating historical energy consumption data and environmental variables to predict energy demand, it can reflect changes in energy demand in real time and adjust energy allocation strategies adaptively. This combination of real-time and adaptability enables the system to always maintain optimal energy efficiency and cost-effectiveness under different environmental conditions and changes in consumer demand. The Markov decision model is used to dynamically allocate charging and discharging priorities and adjust the power allocation strategy. Through the combination of target power allocation strategies, the relationship between energy supply and actual household demand can be more accurately matched, energy waste can be effectively reduced, and the service life of energy storage equipment can be optimized. Through the characteristic analysis of electrical signal data and intelligent strategy adjustment, this method can improve the system's response speed and accuracy to external changes, thereby enhancing the overall stability of the system. Whether it is environmental conditions such as weather changes or fluctuations in the internal load of the home, the system can respond quickly to ensure the stability of power supply. It can effectively utilize renewable energy such as solar and wind energy and reduce dependence on traditional power grids. Through precise electrical signal characteristic analysis and target energy demand prediction, this method can more effectively utilize each unit of energy. Optimizing the charging and discharging process and time ensures sufficient energy when needed while avoiding overcharging or discharging, thereby improving overall energy efficiency, thereby improving the accuracy of power distribution based on energy storage devices and improving the stability of household electricity consumption.

[0082] Figure 3It is a structural schematic diagram of an energy distribution device based on an energy storage device provided by an embodiment of the present invention. The energy distribution device 300 based on the energy storage device may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage device terminals) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 can be short-term storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the energy distribution device 300 based on the energy storage device. Furthermore, the processor 310 may be configured to communicate with the storage medium 330, and execute a series of instruction operations in the storage medium 330 on the energy distribution device 300 based on the energy storage device to implement the steps of the above-mentioned energy distribution method based on the energy storage device.

[0083] The energy distribution device 300 based on the energy storage device may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It can be understood by those skilled in the art that Figure 3 The structure of the energy storage device-based electric energy distribution device shown does not constitute a limitation on the energy storage device-based electric energy distribution device provided by the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0084] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the electric energy distribution method based on the energy storage device.

[0085] 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.

[0086] 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 invention is essentially or the part that contributes to the prior art or the whole 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, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0087] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. 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 invention.

Claims

1. A method for distributing electric energy based on energy storage equipment, characterized in that: The method comprises: Initializing energy management of a hybrid energy storage device to obtain an initial electric energy distribution strategy combination, wherein the hybrid energy storage device includes: a first energy storage device and a second energy storage device; Performing an electric signal data feature analysis on the hybrid energy storage device based on the initial electric energy allocation strategy combination to obtain a first electric signal feature set of the first energy storage device and a second electric signal feature set of the second energy storage device; Acquire historical energy consumption data and environmental variables, and perform energy demand forecasting on the hybrid energy storage device to obtain target energy demand forecasting data; specifically include: perform energy consumption analysis on the hybrid energy storage device to obtain historical energy consumption data, and perform environmental variable monitoring on the hybrid energy storage device to obtain environmental variables; perform vector coding mapping on the historical energy consumption data to obtain an energy consumption coding vector; perform vector fusion on the energy consumption coding vector based on the environmental variables to obtain a target fusion coding vector; input the target fusion coding vector into a preset double-layer LSTM model; perform hidden feature extraction on the target fusion coding vector through the forward LSTM network in the double-layer LSTM model to obtain a forward hidden feature vector; perform hidden feature extraction on the forward hidden feature vector through the backward LSTM network in the double-layer LSTM model to obtain a backward hidden feature vector; perform vector splicing on the forward hidden feature vector and the backward hidden feature vector to obtain a spliced ​​hidden feature vector; perform energy demand forecasting on the spliced ​​hidden feature vector through the fully connected layer in the double-layer LSTM model to obtain target energy demand forecasting data; The first electric signal feature set, the second electric signal feature set and the target energy demand forecast data are input into a preset Markov decision model to perform charge and discharge priority allocation and electric energy distribution strategy adjustment to generate a target electric energy distribution strategy combination.

2. The method for distributing electric energy based on energy storage equipment according to claim 1, characterized in that: The energy management of the hybrid energy storage device is initialized to obtain an initial electric energy distribution strategy combination, wherein the hybrid energy storage device includes: a first energy storage device and a second energy storage device, including: Defining a basic parameter space of a hybrid energy storage device, wherein the basic parameter space includes a plurality of charge and discharge parameters and corresponding value ranges, and the plurality of charge and discharge parameters include: charge and discharge current, charge and discharge voltage, and charge and discharge time period; Generate uniformly distributed random values ​​for each charge and discharge parameter in the basic parameter space by using a sine-cosine algorithm; Generate a first initial electric energy distribution strategy for a first energy storage device and a second initial electric energy distribution strategy for a second energy storage device in the hybrid energy storage device according to the random value of each charging and discharging parameter; According to the preset parameter search space and the number of iterations, a strategy fusion analysis is performed on the first initial electric energy allocation strategy and the second initial electric energy allocation strategy to obtain an initial electric energy allocation strategy combination.

3. The method for distributing electric energy based on energy storage equipment according to claim 1, characterized in that: The step of performing electric signal data feature analysis on the hybrid energy storage device based on the initial electric energy allocation strategy combination to obtain a first electric signal feature set of the first energy storage device and a second electric signal feature set of the second energy storage device includes: Performing energy supply testing and data collection on the hybrid energy storage device based on the initial electric energy distribution strategy combination to obtain first electrical signal data of the first energy storage device and second electrical signal data of the second energy storage device; constructing a first state matrix of the first energy storage device based on the first electrical signal data, and constructing a second state matrix of the second energy storage device based on the second electrical signal data; Performing eigenvalue calculation on the first state matrix to obtain a first eigendifference value, and performing eigenvalue calculation on the second state matrix to obtain a second eigendifference value; The first electrical signal data is subjected to signal reorganization and signal feature extraction according to the first feature difference value to obtain a first electrical signal feature set of the first energy storage device, and the second electrical signal data is subjected to signal reorganization and signal feature extraction according to the second feature difference value to obtain a second electrical signal feature set of the second energy storage device.

4. The method for distributing electric energy based on energy storage equipment according to claim 3, characterized in that: The performing eigenvalue calculation on the first state matrix to obtain a first eigendifference value, and performing eigenvalue calculation on the second state matrix to obtain a second eigendifference value, comprises: Performing eigenvalue calculation on the first state matrix based on the QR algorithm to obtain a plurality of first matrix eigenvalues, and performing eigenvalue calculation on the second state matrix based on the QR algorithm to obtain a plurality of second matrix eigenvalues; Performing maximum value identification on the plurality of first matrix eigenvalues ​​to obtain a first maximum eigenvalue, and performing maximum value identification on the plurality of second matrix eigenvalues ​​to obtain a second maximum eigenvalue; Performing geometric mean calculation on the plurality of first matrix eigenvalues ​​to obtain a first characteristic geometric mean value, and performing geometric mean calculation on the plurality of second matrix eigenvalues ​​to obtain a second characteristic geometric mean value; The first maximum eigenvalue and the first characteristic geometric mean are difference calculated to obtain a first characteristic difference value, and the second maximum eigenvalue and the second characteristic geometric mean are difference calculated to obtain a second characteristic difference value.

5. The method for distributing electric energy based on energy storage equipment according to claim 1, characterized in that: The step of inputting the first electrical signal feature set, the second electrical signal feature set and the target energy demand forecast data into a preset Markov decision model to perform charge and discharge priority allocation and electric energy allocation strategy adjustment to generate a target electric energy allocation strategy combination includes: Creating a first agent of the first energy storage device and a second agent of the second energy storage device, and defining states, actions, and reward functions of Markov decision models respectively; Performing a two-dimensional matrix construction on the first electrical signal feature set and the target energy demand forecast data to obtain a first two-dimensional feature matrix, and performing a two-dimensional matrix construction on the second electrical signal feature set and the target energy demand forecast data to obtain a second two-dimensional feature matrix; Importing the first agent and inputting the first two-dimensional feature matrix into the Markov decision model to perform an electric energy distribution strategy analysis to obtain a first target electric energy distribution strategy, and at the same time, importing the second agent and inputting the second two-dimensional feature matrix into the Markov decision model to perform an electric energy distribution strategy analysis to obtain a second target electric energy distribution strategy; The charging and discharging priorities are assigned to the first intelligent agent and the second intelligent agent to obtain the charging and discharging priority information, and the first target power distribution strategy and the second target power distribution strategy are globally adjusted for energy distribution according to the charging and discharging priority information to generate a target power distribution strategy combination.

6. The method for distributing electric energy based on energy storage equipment according to claim 5, characterized in that: The step of importing the first agent and inputting the first two-dimensional feature matrix into the Markov decision model to perform an electric energy distribution strategy analysis to obtain a first target electric energy distribution strategy, and importing the second agent and inputting the second two-dimensional feature matrix into the Markov decision model to perform an electric energy distribution strategy analysis to obtain a second target electric energy distribution strategy, includes: Importing the first agent into the Agent program and inputting the first two-dimensional feature matrix into the Markov decision model to determine the first candidate power distribution strategy learned by the first agent; importing the second agent into the Agent program and inputting the second two-dimensional feature matrix into the Markov decision model to determine the second candidate power distribution strategy learned by the second agent; Determine a first current action of the first agent according to the first candidate power allocation strategy, and send the first current action to the virtual environment for state update to obtain a first updated state; determine a second current action of the second agent according to the second candidate power allocation strategy, and send the second current action to the virtual environment for state update to obtain a second updated state; Calculating a first reward feedback parameter of the first agent according to the first updated state and the reward function, and calculating a second reward feedback parameter of the second agent according to the second updated state and the reward function; Calculating a first strategy action update gradient of the first agent according to the first reward feedback parameter, and calculating a second strategy action update gradient of the second agent according to the second reward feedback parameter; A first target power distribution strategy of the first agent is determined according to the first strategy action update gradient, and a second target power distribution strategy of the second agent is determined according to the second strategy action update gradient.

7. An electric energy distribution device based on energy storage equipment, characterized in that: The electric energy distribution device based on the energy storage device comprises: An initialization module is used to initialize energy management of a hybrid energy storage device to obtain an initial electric energy distribution strategy combination, wherein the hybrid energy storage device includes: a first energy storage device and a second energy storage device; An analysis module, configured to perform an electric signal data feature analysis on the hybrid energy storage device based on the initial electric energy allocation strategy combination to obtain a first electric signal feature set of the first energy storage device and a second electric signal feature set of the second energy storage device; A prediction module is used to obtain historical energy consumption data and environmental variables, and to predict the energy demand of the hybrid energy storage device to obtain target energy demand prediction data; specifically comprising: performing energy consumption analysis on the hybrid energy storage device to obtain historical energy consumption data, and performing environmental variable monitoring on the hybrid energy storage device to obtain environmental variables; performing vector coding mapping on the historical energy consumption data to obtain an energy consumption coding vector; performing vector fusion on the energy consumption coding vector based on the environmental variables to obtain a target fusion coding vector; inputting the target fusion coding vector into a preset double-layer LSTM model; performing hidden feature extraction on the target fusion coding vector through the forward LSTM network in the double-layer LSTM model to obtain a forward hidden feature vector; performing hidden feature extraction on the forward hidden feature vector through the backward LSTM network in the double-layer LSTM model to obtain a backward hidden feature vector; performing vector splicing on the forward hidden feature vector and the backward hidden feature vector to obtain a spliced ​​hidden feature vector; performing energy demand prediction on the spliced ​​hidden feature vector through the fully connected layer in the double-layer LSTM model to obtain target energy demand prediction data; A generation module is used to input the first electrical signal feature set, the second electrical signal feature set and the target energy demand forecast data into a preset Markov decision model to perform charge and discharge priority allocation and electric energy distribution strategy adjustment to generate a target electric energy distribution strategy combination.

8. An electric energy distribution device based on an energy storage device, characterized in that: The electric energy distribution device based on the energy storage device comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the energy storage device-based electric energy distribution device to execute the energy storage device-based electric energy distribution method according to any one of claims 1-6.

9. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the electric energy distribution method based on the energy storage device as described in any one of claims 1 to 6 is implemented.

Images