New energy storage optimization system and method, storage medium and electronic equipment

Abstract

The invention provides a new energy storage optimization system and method, a storage medium and electronic equipment, and relates to the technical field of new energy. The system comprises an energy storage equipment state evaluation module which is used for obtaining real-time operation data of energy storage equipment and evaluating the health state of the energy storage equipment in combination with historical charging and discharging data of the energy storage equipment; the power grid load demand prediction module is used for determining a target power grid load demand through a trained load prediction model based on the power grid load data and the external environment data; and the energy storage equipment scheduling optimization module is used for determining charging and discharging adjustment parameters of the energy storage equipment by utilizing a scheduling optimization model based on the health state of the energy storage equipment and the target power grid load demand, and adjusting the charging rate and the discharging power of the energy storage equipment according to the charging and discharging adjustment parameters. According to the invention, dynamic optimization scheduling is carried out on the energy storage equipment, so that the service life of the energy storage equipment can be prolonged, the energy utilization efficiency of the energy storage system is improved, and the stability of the energy storage system is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of new energy technology, and in particular to a new energy storage optimization system, method, storage medium and electronic equipment. Background Art

[0002] As the penetration rate of renewable energy power generation in the entire power system continues to increase, in order to balance the randomness and intermittency of wind and solar power generation and improve the entire power system's ability to absorb renewable energy, it is necessary to build a large-scale energy storage system that matches the power system.

[0003] Among them, the intermittent and volatile nature of renewable energy sources such as wind and solar energy leads to uncertainty and instability in wind and photovoltaic power generation. Currently, new energy storage systems have become a key technology to alleviate the imbalance between power supply and demand and improve the stability of the power system. However, in actual applications, new energy storage systems in related technologies have problems such as uneven energy distribution, low energy storage efficiency, and limited equipment life.

[0004] Therefore, there is an urgent need to provide a new renewable energy storage optimization system to improve the efficiency of energy management by optimizing scheduling, thereby ensuring the efficient operation of the energy storage system and the long life of the energy storage equipment.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0006] The purpose of the embodiments of the present disclosure is to provide a new energy storage optimization system, method, storage medium and electronic device, thereby at least to a certain extent solving the problems of uneven energy distribution, low energy storage efficiency and limited equipment life in new energy storage systems in related technologies.

[0007] According to a first aspect of an embodiment of the present disclosure, a new energy storage optimization system is provided, comprising: An energy storage device status assessment module, used to obtain real-time operating data of the energy storage device and assess the health status of the energy storage device in combination with historical charge and discharge data of the energy storage device; A power grid load demand prediction module is used to determine the target power grid load demand through a trained load prediction model based on power grid load data and external environment data; The energy storage device scheduling optimization module is used to determine the charge and discharge adjustment parameters of the energy storage device based on the health status of the energy storage device and the target grid load demand using a scheduling optimization model, and adjust the charging rate and discharge power of the energy storage device according to the charge and discharge adjustment parameters.

[0008] In an exemplary embodiment of the present disclosure, the energy storage device status assessment module includes: A first data acquisition submodule is used to acquire real-time operation data and historical charge and discharge data of the energy storage device, wherein the real-time operation data includes voltage, current, and temperature; The first data processing submodule is used to calculate the internal resistance change according to the voltage and current in the real-time operation data; generate a temperature spatiotemporal distribution matrix containing a timestamp and spatial coordinates according to the temperature, perform sliding window segmentation processing on the temperature spatiotemporal distribution matrix, and obtain the temperature gradient change rate by calculating the temperature change gradient of adjacent time windows; perform sliding window segmentation processing on the historical charge and discharge data to generate a voltage-current change rate data set within the time window; A first data analysis submodule is used to establish a charge-discharge cycle number statistical model based on the voltage-current change rate data set, wherein the charge-discharge cycle number statistical model uses a cubic exponential smoothing method to update the cumulative value of the cycle number; and to match the temperature gradient change rate with the voltage-current change rate data set to generate a temperature-cycle correlation coefficient matrix; The health status assessment submodule is used to fuse the internal resistance change, the cumulative value of the number of cycles and the temperature-cycle correlation coefficient matrix to generate health status assessment parameters, and determine the health status of the energy storage device according to the health status assessment parameters.

[0009] In an exemplary embodiment of the present disclosure, the power grid load demand prediction module includes: The second data acquisition submodule is used to acquire current grid load data, historical load time series data and external environment data; A second data processing submodule is used to perform sliding window segmentation processing on the historical load time series data to generate a load segment data set with time continuity, and to perform principal component analysis and dimensionality reduction processing on the external environment data to generate an environment feature vector set; The multi-dimensional fusion prediction submodule is used to input the current power grid load data, load segment data set and environmental feature vector set into the trained load prediction model, and output the target power grid load demand.

[0010] In an exemplary embodiment of the present disclosure, the load prediction model includes a first load prediction model and a second load prediction model, the target power grid load demand includes a first power grid load demand and a second power grid load demand; the multi-dimensional fusion prediction submodule includes: A feature extraction unit, configured to generate an enhanced short-term feature vector according to the current power grid load data and the load segment data set; A trend decomposition unit, used for performing seasonal trend decomposition on the environmental feature vector set to obtain a long-term trend component and a periodic component, and generating a trend feature matrix according to the long-term trend component and the periodic component; A short-term prediction unit, used for inputting the enhanced short-term feature vector into a trained first load prediction model to obtain a first power grid load demand; The long-term prediction unit is used to input the enhanced short-term feature vector and the trend feature matrix into the trained second load prediction model to obtain the second power grid load demand.

[0011] In an exemplary embodiment of the present disclosure, the energy storage device scheduling optimization module includes: A parameter determination submodule is used to construct a state space set based on the health state of the energy storage device and the target grid load demand, define an action space set according to the charge and discharge adjustment parameters, construct a reward function according to the load matching reward, the health state reward and the energy efficiency reward, calculate the reward values ​​of different actions in different states using the trained scheduling optimization model, and determine the charge and discharge adjustment parameters corresponding to the target reward value; The dynamic scheduling submodule is used to adjust the charging rate and discharging power of the energy storage device according to the charging and discharging adjustment parameters.

[0012] In an exemplary embodiment of the present disclosure, the system further includes: A data feedback module, used to establish an operation data queue including a charge and discharge operation timestamp and an actual power deviation value according to the adjusted charging rate and discharge power of the energy storage device and the operation log data of the energy storage device; A model update module is used to calculate the statistical error between the actual power deviation value and the target power grid load demand based on the charging and discharging operation timestamps in the operating data queue, and trigger the online update operation of the scheduling optimization model through the statistical error; wherein the online update operation includes using the time difference method to calculate the expected cumulative reward function gradient and correct the model parameters of the scheduling optimization model.

[0013] According to a second aspect of an embodiment of the present disclosure, a new energy storage optimization method is provided, comprising: Acquire real-time operating data of the energy storage device, and evaluate the health status of the energy storage device in combination with historical charge and discharge data of the energy storage device; Based on the grid load data and external environment data, the target grid load demand is determined through the trained load prediction model; Based on the health status of the energy storage device and the target grid load demand, a scheduling optimization model is used to determine the charge and discharge adjustment parameters of the energy storage device, and the charging rate and discharge power of the energy storage device are adjusted according to the charge and discharge adjustment parameters to match the real-time load demand of the grid.

[0014] In an exemplary embodiment of the present disclosure, the method further includes: According to the adjusted charging rate and discharging power of the energy storage device and the operation log data of the energy storage device, an operation data queue including a charging and discharging operation timestamp and an actual power deviation value is established; Based on the charging and discharging operation timestamps in the operation data queue, a statistical error between the actual power deviation value and the target power grid load demand is calculated, and model parameters of the scheduling optimization model are updated according to the statistical error.

[0015] According to a third aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, any step of the new energy storage optimization method described in the first aspect is implemented.

[0016] According to a fourth aspect of the present disclosure, there is provided an electronic device, including: A processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any step of the new energy storage optimization method described in the first aspect is implemented.

[0017] The technical solution provided by the embodiments of the present disclosure may have the following beneficial effects: In the new energy storage optimization system provided in the exemplary embodiment of the present disclosure, by evaluating the health status of the energy storage device in real time, it can help the energy storage system optimize the charging rate and discharge power, avoid excessive charging or discharging causing equipment aging, and thus extend the life of the energy storage device; the grid load demand prediction module can grasp the trend of grid load demand changes in advance, so that the energy storage system can adjust the load overload strategy in advance to avoid grid load overload or energy waste; the energy storage device scheduling optimization module can timely adjust the charging and discharging timing to ensure that the energy storage device is charged and discharged in the most optimized efficiency and least charging manner, reducing energy loss. Dynamically adjusting the charging and discharging strategy can not only respond to changes in grid load, but also intelligently adjust charging and discharging according to the health status of the energy storage device, reasonably allocate charging and discharging tasks, avoid excessive pressure on a certain energy storage device and cause equipment aging or failure, and thus improve the energy storage efficiency of the energy storage system. In general, through accurate health status assessment, grid load prediction and intelligent scheduling optimization, the energy efficiency of the energy storage system and the continuous operation of the equipment are improved, the reasonable configuration and effective use of the energy storage system are guaranteed, and the stability and flexibility of the grid are enhanced. By reducing energy waste and optimizing grid efficiency, the economy and reliability of the energy storage system are improved, ensuring that the new energy system can operate more stably when the grid load changes uncertainly.

[0018] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0020] Figure 1 A schematic diagram of the architecture of a new energy storage optimization system in an embodiment of the present disclosure is shown.

[0021] Figure 2 A schematic diagram of the architecture of another new energy storage optimization system in an embodiment of the present disclosure is shown.

[0022] Figure 3 A schematic diagram of a process flow of a new energy storage optimization method in an embodiment of the present disclosure is shown.

[0023] Figure 4 A schematic diagram of the structure of an electronic device suitable for implementing the embodiments of the present disclosure is shown.

[0024] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0025] The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit this specification. The singular forms "a", "the" and "the" used in this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0026] It should be understood that although the terms first, second, third, etc. may be used in this specification to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this specification, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0027] Figure 1 FIG. 1 is a schematic diagram showing the architecture of a new energy storage optimization system according to an embodiment of the present disclosure. Figure 1 As shown, the new energy storage optimization system 100 includes an energy storage device state evaluation module 110, a power grid load demand prediction module 120 and an energy storage device scheduling optimization module 130. Among them, the energy storage device state evaluation module 110 is used to obtain the real-time operation data of the energy storage device, and evaluate the health status of the energy storage device in combination with the historical charging and discharging data of the energy storage device. The power grid load demand prediction module 120 is used to determine the target power grid load demand through a trained load prediction model based on the power grid load data and external environment data. The energy storage device scheduling optimization module 130 is used to determine the charge and discharge adjustment parameters of the energy storage device based on the health status of the energy storage device and the target power grid load demand, using the scheduling optimization model, and adjust the charging rate and discharge power of the energy storage device according to the charge and discharge adjustment parameters.

[0028] This disclosure improves the energy efficiency of the energy storage system and the continuous operation of the equipment through accurate health status assessment, grid load prediction and intelligent dispatch optimization, ensures the reasonable configuration and effective use of the energy storage system, and enhances the stability and flexibility of the grid. By reducing energy waste and optimizing grid efficiency, the economy and reliability of the energy storage system are improved, ensuring that the new energy system can operate more stably when the grid load changes uncertainly.

[0029] refer to Figure 1As shown, the energy storage device state assessment module 110 includes a first data acquisition submodule 111, a first data processing submodule 112, a first data analysis submodule 113 and a health state assessment submodule 114. Among them, the first data acquisition submodule 111 is used as the data entry of the energy storage device state assessment module 110 to collect the real-time operation data and historical charge and discharge data of the energy storage device to build the original data basis for the assessment. In the embodiment of the present disclosure, the energy storage device can be a battery pack, a super capacitor, etc. Taking the battery pack as an example, the real-time operation data of the energy storage device includes but is not limited to the battery's power, temperature, charge and discharge cycle, voltage, current, internal resistance and other health indicators. The historical charge and discharge data of the energy storage device includes but is not limited to charge and discharge curve data, charge and discharge capacity data, charge and discharge cycle data, charge and discharge rate data, charge and discharge temperature data, and charge and discharge time data. The historical charge and discharge data of the energy storage device can reflect the operating status, performance change and aging trend of the energy storage device, and is the key input of the energy storage device state assessment module 110.

[0030] Exemplarily, the first data acquisition submodule 111 acquires the three key parameters of the battery, namely, voltage, current and temperature, in real time, and acquires the voltage-current-time curve of the charge-discharge cycle from the historical database. For example, the battery management system provides the voltage and current of the battery in real time, and the distributed temperature sensor acquires the temperature parameters of the battery in real time. Among them, the voltage of the battery can directly reflect the charging state and health level of the battery. By measuring the current data of the battery during the charging and discharging process, it is helpful to determine whether the battery is in normal working condition. Temperature monitoring is crucial to the battery. For example, too high a temperature will accelerate the aging of the battery, while too low a temperature may cause the battery performance to deteriorate.

[0031] The first data processing submodule 112 is used to process the original data, such as the voltage, current and temperature of the battery, to generate three types of core characteristic indicators, namely, the internal resistance change, the temperature gradient change rate, and the voltage-current change rate data set, to provide support for subsequent evaluation.

[0032] In an example implementation, in the first data processing submodule 112, the internal resistance change can be calculated based on the voltage and current in the real-time operation data, and the change in internal resistance reflects the degree of aging or damage of the energy storage device. A temperature spatiotemporal distribution matrix containing a timestamp and spatial coordinates is generated based on the temperature, and the temperature spatiotemporal distribution matrix is ​​subjected to sliding window segmentation processing. The temperature gradient of adjacent time windows is calculated to obtain the temperature gradient change rate. The historical charge and discharge data is subjected to sliding window segmentation processing to generate a voltage-current change rate data set within the time window.

[0033] Specifically, the current internal resistance value can be calculated based on the voltage and current in the real-time operation data, and the ratio of the current internal resistance value to the initial internal resistance value can be used as the internal resistance change. The temperature data not only contains the instantaneous temperature of the energy storage device, but may also change over time and space. Therefore, the temperature spatiotemporal distribution matrix can be constructed based on the temperature in the real-time operation data, which is recorded as , where N is the number of temperature sensors, M is the number of time slices (such as 10 seconds / slice), and K is the spatial coordinate dimension. Next, a sliding window is used to segment the temperature spatiotemporal distribution matrix to analyze the temperature changes in different time periods. For example, the window length is 5 minutes, the step length is 30 seconds, and the temperature gradient within the window is calculated: (1) in, for t The temperature gradient at time, is the spatial weight coefficient, Express Perform spatial gradient calculation. By performing differential operation on the gradients of adjacent windows, the temperature gradient change rate can be obtained, which is recorded as , thereby evaluating the impact of temperature fluctuations on the health status of energy storage equipment.

[0034] After obtaining historical charge and discharge data such as the voltage-current-time curve of the charge and discharge cycle, it can be divided into time windows, such as a window length of 10% of a single cycle. Of course, the window length can be adjusted adaptively. The voltage change rate and current change rate are calculated in each window to form a voltage-current change data set, which is recorded as ,The voltage change rate and current change rate are the key factors for evaluating whether the charging and discharging behavior of the energy storage equipment is normal.

[0035] The first data processing submodule 112 realizes a multi-dimensional dynamic analysis of the operating status of the energy storage device through dynamic internal resistance monitoring and temperature spatiotemporal distribution analysis. The sliding window segmentation and gradient calculation of the temperature parameters can accurately capture the transient characteristics of charging and discharging and local temperature anomalies. Combined with the dynamic calibration of the voltage-current change rate data set, it can significantly improve the detection sensitivity of internal resistance changes and thermal anomalies, provide high-precision and high-efficiency pre-processing data support for subsequent health assessments, and effectively enhance the safety monitoring capability and life prediction reliability of the energy storage system.

[0036] Furthermore, the first data analysis submodule 113 is used to establish a charge-discharge cycle number statistical model based on the voltage-current change rate data set, wherein the charge-discharge cycle number statistical model uses a cubic exponential smoothing method to update the cumulative value of the cycle number. The model can reflect the frequency of use of the energy storage device and evaluate the aging degree of the energy storage device based on the accumulated number of cycles. It should be noted that a cumulative discharge depth of 80% can be regarded as a complete cycle.

[0037] For example, when the triple exponential smoothing method is used to update the cumulative value of the number of cycles, there are: (2) in, is the triple smoothed value at time t, is the triple smoothed value at time t-1, is the smoothing coefficient, is the equivalent number of cycles in the current window, is a trend term used to quantify the growth rate of the number of cycles, such as When the charge and discharge frequency is accelerated, the aging is accelerated. When , it indicates that the cycle number region is stable; is a seasonal term used to capture the superposition effect of periodic charge and discharge patterns on cycle aging, such as the surge in daily cycle times caused by air conditioning load in summer.

[0038] Formula (2) can more accurately reflect the actual degree of aging by converting the discharge depth based on the number of cycles and correcting it with the trend term / seasonal term.

[0039] The first data analysis submodule 113 also needs to match the temperature gradient change rate with the voltage-current change rate data set to generate a temperature-cycle correlation coefficient matrix. With the voltage-current variation data set The joint data set is formed by aligning the timestamps. Then, the temperature gradient change rate of each spatial position in each time window is used as the temperature feature, and the voltage-current change data set is obtained. The voltage change rate and current change rate are extracted as cycle features. For each temperature spatial position, the sliding window covariance between the corresponding temperature gradient change rate and the voltage change rate or current change rate is calculated, and then the correlation coefficient is obtained by normalization. Finally, the temperature-cycle correlation coefficient matrix is ​​composed of the temperature sensor position (spatial dimension), cycle features and correlation coefficients. The temperature-cycle correlation coefficient matrix can be used to locate temperature-sensitive areas during charging and discharging, optimize thermal management strategies or warn of abnormal operating conditions.

[0040] The first data analysis submodule 113 establishes a statistical model of the number of charge and discharge current cycles by analyzing the voltage-current change rate data set. The model is updated using the cubic exponential smoothing method to help accurately track the number of cycle relationships of the energy storage device and indirectly reflect the use of the energy storage device. In addition, by matching the temperature gradient change rate and the voltage-current change rate data set, a temperature-cycle correlation coefficient matrix is ​​constructed to quantify the relationship between temperature fluctuations and charge and discharge current cycles, providing more data support for health status assessment.

[0041] The health status assessment submodule 114 is responsible for integrating and analyzing the internal resistance change, the cumulative value of the number of charge and discharge cycles, and the temperature-cycle correlation coefficient matrix to generate health status assessment parameters, thereby comprehensively judging the overall health status of the energy storage device.

[0042] For example, the internal resistance change, the cumulative value of the number of charge and discharge cycles, and the temperature-cycle correlation coefficient matrix can be normalized respectively. For example, the internal resistance change is converted into a percentage form and recorded as , normalize the accumulated value of the number of cycles according to the rated number of cycles, and record it as , the temperature-cycle correlation coefficient matrix is ​​taken as the temperature influence factor, which is recorded as Then, the normalized internal resistance change, the cumulative value of the charge and discharge cycles, and the temperature-cycle correlation coefficient matrix are weighted and fused to obtain: (3) in, for health status assessment parameters, is the weight coefficient, is the temperature attenuation coefficient.

[0043] For example, if S≥80%, it means that the energy storage device is in a healthy state; if 60%≤S<80%, it means that the energy storage device is in a warning state; if S<60%, the energy storage device is in a failure risk state.

[0044] The energy storage device status assessment module 110 in the embodiment of the present disclosure is a multi-level data-driven analysis system, which realizes a dynamic quantitative assessment of the health status of the energy storage device by integrating real-time operation data with historical data, combining spatiotemporal feature modeling, cycle count statistics and cross-dimensional correlation analysis.

[0045] The power grid load demand prediction module 120 includes a second data acquisition submodule 121, a second data processing submodule 122 and a multi-dimensional fusion prediction submodule 123. Among them, the second data acquisition submodule 121, as the data input end of the power grid load demand prediction module 120, is responsible for the real-time collection and preliminary cleaning of multi-source heterogeneous data, and provides an original data pool for downstream processing. Exemplarily, the second data acquisition submodule 121 is used to obtain relevant data from the power grid system and the external environment, including current power grid load data, historical load time series data, and external environment data that affects load changes. The current power grid load data is the power demand of the power grid at the current moment, which can be collected in real time by smart meters, monitoring and data acquisition systems or load management systems. Historical load time series data refers to the power grid load data of the past period of time, usually including hourly, daily or longer time span data. Trend analysis of historical load time series data can identify periodic fluctuations and sudden demand changes. External environmental data include meteorological factors, time factors and socio-economic data. Meteorological factors include temperature, humidity, wind speed, light intensity, etc. Weather conditions have a significant impact on the load of the power grid. For example, hot weather may increase the use of air conditioners, thereby increasing the load. Time factors include working days, holidays, seasonal changes, etc. Load demand may have regularity at different time nodes. For example, weekend load is usually lower than weekdays. Socio-economic data include industrial production levels, residents' electricity consumption habits, etc. Socio-economic activities also have a certain impact on the change of power grid load. External environmental data can be used as an auxiliary input of the model to optimize the accuracy of load demand prediction.

[0046] It can be understood that after obtaining the current power grid load data, historical load time series data and external environment data, data preprocessing can be performed, for example, outlier filtering and missing value interpolation are performed on the real-time power grid load data, time alignment and unit normalization are performed on the historical load time series data, field merging and timestamp matching are performed on the external environment data, and structured table data is generated. The present disclosure does not limit this.

[0047] The second data processing submodule 122 is used to perform structured transformation and feature engineering on the original data, generate time-continuous load segments and dimension-reduced environmental features, and provide adaptive input for the prediction model.

[0048] Specifically, in order to capture the changing trend of the power grid load, the historical load time series data is first subjected to sliding window segmentation processing to generate a load segment data set with time continuity. For example, the sliding window length T=24 (6 hours*4) and the sliding step length S=4 (1 hour) are defined, and the historical load time series data is divided into multiple continuous load segments according to the window size. Gaussian noise and random time offset (±5 minutes) can also be added to the window data to improve the robustness of the model. Sliding window segmentation can retain the continuity of time series data, enabling the model to learn the short-term and long-term change patterns of the load and improve the prediction accuracy.

[0049] Since the external environmental data contains multiple variables, the data dimension is too high. Directly inputting it into the load forecasting model may increase the computational complexity and introduce noise. Therefore, the external environmental data can be subjected to principal component analysis and dimensionality reduction processing to extract the main features from the high-dimensional environmental data, remove redundant information, and generate an environmental feature vector set. The environmental feature vector after dimensionality reduction can more accurately represent the impact of the external environment on the power grid load, so that the load forecasting model can use environmental factors more efficiently, thereby improving the generalization ability of the forecasting model.

[0050] For example, the N-dimensional external environment data can be spliced ​​into an N×T matrix according to timestamps, retaining 95% variance contribution rate, and the N-dimensional environment data can be compressed to K dimensions (K≤N) to generate an environmental feature vector set.

[0051] The multi-dimensional fusion prediction submodule 123 receives the load segment data set and environmental feature vector set processed by the second data processing submodule, and inputs them into the trained load prediction model in combination with the current grid load data, and finally outputs the target grid load demand. The prediction results will be used to optimize energy storage scheduling and ensure the stability and efficient operation of the power system.

[0052] Exemplarily, the load prediction model can be a time series model, such as a long short-term memory network, Transformer, etc., which is used to capture the time change trend of the power grid load. It can also be a regression model, such as XGBoost, random forest, etc., to combine environmental data and historical loads for regression prediction. It can also be a hybrid model, such as deep learning model + statistical learning, which improves the prediction accuracy by integrating neural networks and statistical methods. The present disclosure does not limit the specific type and network structure of the load prediction model.

[0053] In an example implementation, the load forecasting model includes a first load forecasting model and a second load forecasting model, and the target power grid load demand includes a first power grid load demand and a second power grid load demand. The first load forecasting model can be a deep learning convolutional neural network or a recursive neural network for extracting multi-dimensional short-term load change patterns, and the second load forecasting model can be a Transformer model, a time convolutional network, etc., for capturing long-term trends. Accordingly, the first power grid load demand is the short-term load forecast result, and the second power grid load demand is the long-term load forecast result.

[0054] refer to Figure 1 As shown, the multi-dimensional fusion prediction submodule 123 includes a feature extraction unit 1231, a trend decomposition unit 1232, a short-term prediction unit 1233 and a long-term prediction unit 1234.

[0055] Specifically, the feature extraction unit 1231 is used to generate an enhanced short-term feature vector based on the current power grid load data and the load segment data set. For example, the current power grid load data is subjected to sliding average filtering for noise reduction, and the time domain features within the window, such as mean, variance, etc., are calculated, and the frequency domain features in the current power grid load data are extracted by fast Fourier transform, and the extracted time domain features and frequency domain features are spliced ​​with the load segment data with high similarity in the load segment data set to obtain an enhanced short-term feature vector.

[0056] The trend decomposition unit 1232 is used to perform seasonal trend decomposition on the environmental feature vector set to obtain long-term trend components and periodic components, and generate a trend feature matrix based on the long-term trend components and periodic components. Optionally, STL (Seasonal-Trend Decomposition using LOESS, trend, seasonality and residual) decomposition is used to filter out short-term fluctuations through LOESS smoothing (such as window = 365 days), retain the annual / quarterly level trends, and obtain the long-term trend component. The Fourier series fitting of a fixed period (such as week, month) separates the periodic fluctuations to obtain the periodic component, and the remaining residual part is regarded as random noise and does not participate in subsequent processing. Finally, the long-term trend component and the periodic component are spliced ​​into a trend feature matrix according to the timestamp. The trend feature matrix can provide trend information for long-term load forecasting and help the load forecasting model capture the long-term change trend of the power grid load.

[0057] The short-term prediction unit 1233 is used to input the enhanced short-term feature vector into the trained first load prediction model to obtain the first power grid load demand. The first load prediction model focuses on the power grid load changes within a short time range (such as the next few hours) and is used for real-time scheduling.

[0058] For example, the short-term prediction unit 1233 adopts a temporal convolutional network architecture, with the input being the enhanced short-term feature vector generated by the feature extraction unit 1231. It captures temporal dependencies through multiple layers of causal dilation convolutions, and finally outputs the load demand for the next 15 minutes through a fully connected layer. The smooth L1 loss function and Adam optimizer are used in model training, combined with an early stopping mechanism to prevent overfitting.

[0059] The long-term prediction unit 1234 is used to input the enhanced short-term feature vector and the trend feature matrix into the trained second load prediction model to obtain the second power grid load demand. The second load prediction model focuses on the load changes in a longer time range (such as the next day or a longer period), which helps to plan the charging and discharging strategy of the energy storage system and ensure that the power system can be optimized and dispatched in advance.

[0060] For example, the long-term prediction unit 1234 is based on the Transformer-Encoder structure, and reduces the dimension of the enhanced short-term feature vector through linear projection, and inputs it together with the trend feature matrix output by the trend decomposition unit 1232 and processed by position encoding into a multi-layer self-attention encoder. The feature alignment of short-term dynamics and macro trends is achieved through the cross-attention mechanism, and finally the future 24-hour load demand sequence is generated by full connection. During the model training process, a multi-step mean square error loss function and a RAdam optimizer are used, and gradient clipping is applied to ensure training stability. The first load prediction model and the second load prediction model realize information interaction by sharing short-term features, but the model parameters are updated independently to meet the prediction accuracy requirements of different time scales.

[0061] The coordinated work of the various submodules in the grid load demand prediction module 120 can improve the accuracy of grid load prediction, provide key support for the dispatch optimization and energy storage management of the power system, and ensure that the energy storage system can dispatch energy in a timely manner according to the fluctuating demand of the grid.

[0062] The energy storage device scheduling optimization module 130 includes a parameter determination submodule 131 and a dynamic scheduling submodule 132. The parameter determination submodule 131 serves as the optimization decision core, wherein the scheduling optimization model may adopt a reinforcement learning algorithm, a genetic algorithm, a particle swarm optimization algorithm, etc., which is not limited in the present disclosure. For example, a state-action-reward mapping relationship is constructed through a reinforcement learning framework to generate optimal charging and discharging strategy parameters. The dynamic scheduling submodule 132 serves as the execution end, and converts the charging and discharging parameters output by the parameter determination submodule 131 into real-time control instructions for the energy storage device to achieve dynamic power regulation. The parameter determination submodule 131 and the dynamic scheduling submodule 132 work together, and the energy storage device can efficiently meet the needs of the power grid while ensuring a healthy state, thereby improving the overall energy efficiency and operational stability of the energy storage system.

[0063] Exemplarily, the parameter determination submodule 131 determines the optimal charging and discharging adjustment parameters by using the scheduling optimization model through state space construction, action space definition and reward function calculation. Specifically, a state space set is constructed based on the health status of the energy storage device and the target grid load demand to describe the current operation status of the energy storage device, which is expressed as ,in, S is the health status of the energy storage device, SOC is the real-time power of the energy storage device, is the target grid load demand, and t is the timestamp. Define the action space set according to the charge and discharge adjustment parameters, which is used to define the charge and discharge strategy that the energy storage device can execute, expressed as A={(charging rate, discharge power)}. Construct a reward function based on the load matching reward, health status reward and energy efficiency reward to guide the scheduling optimization model to learn the optimal strategy. It can be understood that the smaller the deviation between the discharge power and the demand, the higher the load matching reward, the higher the charging rate, the lower the health status reward, the higher the discharge efficiency, and the higher the energy efficiency reward. Use the trained scheduling optimization model to calculate the reward values ​​of different actions in different states, and determine the charge and discharge adjustment parameters corresponding to the target reward value, that is, to determine the optimal charge and discharge strategy.

[0064] The dynamic scheduling submodule 132 is used to adjust the charging rate and discharge power of the energy storage device according to the charging and discharging adjustment parameters. In the case of changes in load demand or updates to the health status of the equipment, the dynamic scheduling submodule 132 can adaptively adjust the charging and discharging strategy to ensure that the energy storage device meets the grid demand in the optimal operating state. For example, according to the target grid load demand, the charging rate is dynamically controlled, including adopting a fast charging strategy during the load valley period; when the load fluctuates greatly, a smooth charging strategy is adopted to reduce the fluctuation of the charge state of the energy storage device. For another example, during the peak load period, an appropriate amount of discharge power is provided according to the target load demand to ensure the stability of the grid; when the health status of the energy storage device decreases, the discharge power is reduced to avoid excessive loss of the equipment. For another example, if the grid load suddenly increases, the discharge power is increased; if the temperature of the energy storage device is too high, the charging and discharging rate is reduced; if the health status of the equipment deteriorates, the system adopts a gentle charging and discharging strategy to extend the life of the equipment.

[0065] refer to Figure 2 As shown, a schematic diagram of the architecture of another new energy storage optimization system according to an embodiment of the present disclosure is shown. Figure 2In the new energy storage optimization system 100, a data feedback module 140 and a model update module 150 are also included. The data feedback module 140 is used to align the adjusted charging rate and discharge power of the energy storage device and the operation log data of the energy storage device according to the timestamp, and calculate the actual power deviation value according to the target load demand and discharge power, and then store the charging and discharging operation timestamp and the actual power deviation value in the operation data queue for subsequent use in updating the scheduling optimization model.

[0066] The model update module 150 is used to calculate the statistical error between the actual power deviation value and the target power grid load demand based on the charging and discharging operation timestamps in the operation data queue, and trigger the online update operation of the scheduling optimization model through the statistical error. For example, the mean square error between the two is calculated, and if the mean square error exceeds a preset threshold or the continuous upward trend lasts for T minutes (set according to actual needs), the online update operation is started.

[0067] Among them, the online update operation includes using the time difference method to calculate the expected cumulative reward function gradient and correcting the model parameters of the scheduling optimization model, so that the energy storage system can adapt to new load changes and state changes of energy storage equipment, thereby improving the accuracy of scheduling decisions.

[0068] Specifically, for the state-action pairs in the running data queue ( S t , a t ), calculate the time difference error ,have: (4) in, For the current reward, is the discount factor, Q is the value function. Then, according to the time difference error Adjust the model parameters of the scheduling optimization model, including: (5) in, are the updated model parameters, are the current model parameters, is the learning rate, represents the expected cumulative reward function gradient.

[0069] The new energy storage optimization system 100 continuously optimizes the dispatching strategy through a feedback mechanism, and continuously improves the dispatching optimization model using algorithms such as reinforcement learning, thereby improving the accuracy and adaptability of dispatching decisions.

[0070] The exemplary embodiment of the present disclosure also provides a new energy storage optimization method, which is implemented based on the new energy storage optimization system. Figure 3 As shown, the method may include steps S310 to S330: Step S310, acquiring real-time operation data of the energy storage device, and evaluating the health status of the energy storage device in combination with historical charge and discharge data of the energy storage device.

[0071] Among them, the real-time operation data of the energy storage device, including key parameters such as voltage, current, and temperature, is obtained, and analyzed in combination with historical charge and discharge data to evaluate the health status of the energy storage device. For example, the change in internal resistance of the battery is calculated based on the voltage and current of the battery to determine the degree of battery aging. The temperature spatiotemporal distribution matrix is ​​constructed based on the collected temperature data, and the temperature spatiotemporal distribution matrix is ​​subjected to sliding window segmentation processing. The temperature change gradient of adjacent time windows is calculated to obtain the temperature gradient change rate to analyze the thermal characteristics and temperature change trend of the energy storage device. The historical charge and discharge data is subjected to sliding window processing, the voltage-current change rate data set is extracted, and a statistical model of the number of charge and discharge cycles is established in combination with the cubic exponential smoothing method. Finally, the change in internal resistance, the cumulative value of the number of cycles, and the temperature gradient change rate are fused to generate health status assessment parameters to comprehensively judge the health status of the energy storage device.

[0072] Step S320, based on the grid load data and the external environment data, determine the target grid load demand through the trained load prediction model.

[0073] First, obtain the current power grid load data, historical load time series data and external environment data, such as temperature, humidity, holidays, etc. Then, perform sliding window segmentation on the historical load time series data to generate a time-continuous load segment data set, and perform principal component analysis and dimensionality reduction on the external environment data to extract key environmental features. Finally, input the current power grid load data, load segment data set and environmental feature vector set into the trained load prediction model, calculate the target power grid load demand through time series analysis or deep learning methods, and provide data support for energy storage scheduling. It should be noted that the specific process of calculating the target power grid load demand can refer to the detailed description of the power grid load demand prediction module 120, which will not be repeated here.

[0074] Step S330, based on the health status of the energy storage device and the target grid load demand, use the scheduling optimization model to determine the charge and discharge adjustment parameters of the energy storage device, and adjust the charging rate and discharge power of the energy storage device according to the charge and discharge adjustment parameters to match the real-time load demand of the grid.

[0075] For example, the operation of the energy storage device can be described by constructing a state space set, such as the real-time power of the energy storage device, battery temperature, load trend, etc., and an executable charging and discharging strategy can be defined. Secondly, a reward function is constructed based on load matching, health status, and energy efficiency, and the optimality of different charging and discharging strategies is calculated through optimization algorithms such as reinforcement learning. Finally, the optimal charging and discharging adjustment parameters are determined in combination with the target reward value, and the charging rate and discharge power of the energy storage device are dynamically adjusted to adapt to the fluctuations in the grid load, while taking into account the health status of the equipment to ensure the efficient operation of the energy storage system.

[0076] Similarly, the specific implementation process of step S330 can refer to the detailed description of the energy storage device scheduling optimization module 130, which will not be repeated here.

[0077] Furthermore, according to the adjusted charging rate and discharging power of the energy storage device and the operation log data of the energy storage device, an operation data queue including the charging and discharging operation timestamp and the actual power deviation value is established. Based on the charging and discharging operation timestamps in the operation data queue, the statistical error between the actual power deviation value and the target power grid load demand is calculated, and the model parameters of the scheduling optimization model are updated according to the statistical error.

[0078] Specifically, based on the adjusted charging rate and discharge power of the energy storage device, as well as the operation log data of the energy storage device, an operation data queue is established, which contains the timestamp of each charging and discharging operation and the corresponding actual power deviation value. Among them, the actual power deviation value reflects the difference between the actual output power of the energy storage device and the target grid load demand. Then, based on the timestamp information in the operation data queue, the statistical error of the power deviation in different time periods is calculated to analyze whether the charging and discharging execution of the energy storage device deviates from the predicted target. When the statistical error exceeds the set threshold, the update process of the scheduling optimization model is triggered, the time series analysis method is used to calculate the model adjustment direction, and the optimization algorithm such as the time difference method is used to correct the model parameters, so that the scheduling optimization model can better adapt to the actual operating characteristics of the energy storage device and the dynamic changes of the grid load, thereby improving the scheduling accuracy and load matching capability of the energy storage system.

[0079] The present disclosure realizes the intelligent scheduling of the energy storage system through the full process optimization of energy storage equipment health status assessment, grid load demand prediction, scheduling optimization, data feedback and model update, and improves the operating efficiency of the energy storage equipment and the matching degree of the grid load. By accurately assessing the health status of the energy storage equipment, optimizing the charging and discharging strategy, avoiding overcharging and over-discharging, the life of the energy storage equipment is extended. Based on load prediction and environmental factor analysis, the prediction accuracy of the grid load demand is improved, so that the energy storage system can adjust the charging and discharging plan in advance. The scheduling optimization model dynamically adjusts the charging and discharging rate and power to ensure that the energy storage equipment operates efficiently while meeting the grid demand. The data feedback and model update mechanism enable the system to adaptively adjust the optimization strategy and continuously improve the accuracy of scheduling decisions. The new energy storage optimization method disclosed in the present disclosure improves the energy efficiency and reliability of the energy storage system, reduces the risk of grid load fluctuations, enhances the new energy absorption capacity, and provides an efficient solution for the grid connection of smart grids and renewable energy.

[0080] The exemplary embodiments of the present disclosure also provide a computer-readable storage medium on which a program product capable of implementing the above-mentioned method of the present specification is stored. In some possible implementations, various aspects of the present disclosure may also be implemented in the form of a program product, which includes a program code, and when the program product is run on an electronic device, the program code is used to enable the electronic device to perform the steps according to various exemplary embodiments of the present disclosure described in the above-mentioned "Exemplary Method" section of the present specification.

[0081] The program product may be in the form of a portable compact disk read-only memory (CD-ROM) and include program code, and may be run on an electronic device, such as a personal computer. However, the program product of the present disclosure is not limited thereto, and in this document, a readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0082] The program product may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0083] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0084] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0085] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C#, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0086] In addition, exemplary embodiments of the present disclosure also provide an electronic device capable of implementing the above method.

[0087] Refer to the following Figure 4 hereinafter describes an electronic device 400 according to such an embodiment of the present disclosure. Figure 4 The electronic device 400 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0088] like Figure 4 As shown, the electronic device 400 is in the form of a general computing device. The components of the electronic device 400 may include but are not limited to: the at least one processing unit 410, the at least one storage unit 420, a bus 430 connecting different system components (including the storage unit 420 and the processing unit 410), and a display unit 440.

[0089] The storage unit 420 stores program codes, which can be executed by the processing unit 410, so that the processing unit 410 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification. For example, the processing unit 410 can perform the method steps in the exemplary embodiments of the present disclosure.

[0090] The storage unit 420 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 421 and / or a cache memory unit (Cache) 422 , and may further include a read-only memory unit (ROM) 423 .

[0091] The storage unit 420 may also include a program / utility 424 having a set (at least one) of program modules 425, such program modules 425 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0092] Bus 430 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0093] The electronic device 400 may also communicate with one or more external devices 500 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 400, and / or may communicate with any device that enables the electronic device 400 to communicate with one or more other computing devices (e.g., routers, modems, etc.). This communication may be performed through an input / output (I / O) interface 450. In addition, the electronic device 400 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) through a network adapter 460. As shown, the network adapter 460 communicates with other modules of the electronic device 400 through a bus 430. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 400, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0094] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0095] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present disclosure, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0096] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0097] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

[0098] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A new energy storage optimization system, characterized in that: include: An energy storage device status assessment module, used to obtain real-time operating data of the energy storage device and assess the health status of the energy storage device in combination with historical charge and discharge data of the energy storage device; A power grid load demand prediction module is used to determine the target power grid load demand through a trained load prediction model based on power grid load data and external environment data; The energy storage device scheduling optimization module is used to determine the charge and discharge adjustment parameters of the energy storage device based on the health status of the energy storage device and the target grid load demand using a scheduling optimization model, and adjust the charging rate and discharge power of the energy storage device according to the charge and discharge adjustment parameters.

2. The new energy storage optimization system according to claim 1, characterized in that: The energy storage device status assessment module includes: A first data acquisition submodule is used to acquire real-time operation data and historical charge and discharge data of the energy storage device, wherein the real-time operation data includes voltage, current, and temperature; The first data processing submodule is used to calculate the internal resistance change according to the voltage and current in the real-time operation data; generate a temperature spatiotemporal distribution matrix containing a timestamp and spatial coordinates according to the temperature, perform sliding window segmentation processing on the temperature spatiotemporal distribution matrix, and obtain the temperature gradient change rate by calculating the temperature change gradient of adjacent time windows; perform sliding window segmentation processing on the historical charge and discharge data to generate a voltage-current change rate data set within the time window; A first data analysis submodule is used to establish a charge-discharge cycle number statistical model based on the voltage-current change rate data set, wherein the charge-discharge cycle number statistical model uses a cubic exponential smoothing method to update the cumulative value of the cycle number; and to match the temperature gradient change rate with the voltage-current change rate data set to generate a temperature-cycle correlation coefficient matrix; The health status assessment submodule is used to fuse the internal resistance change, the cumulative value of the number of cycles and the temperature-cycle correlation coefficient matrix to generate health status assessment parameters, and determine the health status of the energy storage device according to the health status assessment parameters.

3. The new energy storage optimization system according to claim 1, characterized in that: The power grid load demand prediction module includes: The second data acquisition submodule is used to acquire current grid load data, historical load time series data and external environment data; A second data processing submodule is used to perform sliding window segmentation processing on the historical load time series data to generate a load segment data set with time continuity, and to perform principal component analysis and dimensionality reduction processing on the external environment data to generate an environment feature vector set; The multi-dimensional fusion prediction submodule is used to input the current power grid load data, load segment data set and environmental feature vector set into the trained load prediction model, and output the target power grid load demand.

4. The new energy storage optimization system according to claim 3 is characterized in that: The load prediction model includes a first load prediction model and a second load prediction model, the target power grid load demand includes a first power grid load demand and a second power grid load demand; the multi-dimensional fusion prediction submodule includes: A feature extraction unit, configured to generate an enhanced short-term feature vector according to the current power grid load data and the load segment data set; A trend decomposition unit, used for performing seasonal trend decomposition on the environmental feature vector set to obtain a long-term trend component and a periodic component, and generating a trend feature matrix according to the long-term trend component and the periodic component; A short-term prediction unit, used for inputting the enhanced short-term feature vector into a trained first load prediction model to obtain a first power grid load demand; The long-term prediction unit is used to input the enhanced short-term feature vector and the trend feature matrix into the trained second load prediction model to obtain the second power grid load demand.

5. The new energy storage optimization system according to claim 1, characterized in that: The energy storage device scheduling optimization module includes: A parameter determination submodule is used to construct a state space set based on the health state of the energy storage device and the target grid load demand, define an action space set according to the charge and discharge adjustment parameters, construct a reward function according to the load matching reward, the health state reward and the energy efficiency reward, calculate the reward values ​​of different actions in different states using the trained scheduling optimization model, and determine the charge and discharge adjustment parameters corresponding to the target reward value; The dynamic scheduling submodule is used to adjust the charging rate and discharging power of the energy storage device according to the charging and discharging adjustment parameters.

6. The new energy storage optimization system according to claim 1, characterized in that: The system further comprises: A data feedback module, used to establish an operation data queue including a charge and discharge operation timestamp and an actual power deviation value according to the adjusted charging rate and discharge power of the energy storage device and the operation log data of the energy storage device; A model update module is used to calculate the statistical error between the actual power deviation value and the target power grid load demand based on the charging and discharging operation timestamps in the operating data queue, and trigger the online update operation of the scheduling optimization model through the statistical error; wherein the online update operation includes using the time difference method to calculate the expected cumulative reward function gradient and correct the model parameters of the scheduling optimization model.

7. A new energy storage optimization method, characterized in that: include: Acquire real-time operating data of the energy storage device, and evaluate the health status of the energy storage device in combination with historical charge and discharge data of the energy storage device; Based on the grid load data and external environment data, the target grid load demand is determined through the trained load prediction model; Based on the health status of the energy storage device and the target grid load demand, a scheduling optimization model is used to determine the charge and discharge adjustment parameters of the energy storage device, and the charging rate and discharge power of the energy storage device are adjusted according to the charge and discharge adjustment parameters to match the real-time load demand of the grid.

8. The new energy storage optimization method according to claim 7, characterized in that: The method further comprises: According to the adjusted charging rate and discharging power of the energy storage device and the operation log data of the energy storage device, an operation data queue including a charging and discharging operation timestamp and an actual power deviation value is established; Based on the charging and discharging operation timestamps in the operation data queue, a statistical error between the actual power deviation value and the target power grid load demand is calculated, and model parameters of the scheduling optimization model are updated according to the statistical error.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the new energy storage optimization method described in any one of claims 7-8 is implemented.

10. An electronic device, characterized in that: include: Processor; and A memory having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by the processor, implement the new energy storage optimization method as described in any one of claims 7-8.

Images