Self-adaptive source-load-storage coordinated optimization control method and system

Abstract

The invention provides a self-adaptive source-load-storage coordinated optimization control method and system, and the method comprises the steps: S1, collecting multi-element load data generated during the operation of a microgrid, and forming a time series data set; s2, key variable characteristics which have the largest influence on the micro-grid load operation state in the micro-grid load requirements are recognized; s3, determining and constructing a source-load-storage cooperative control model based on the key variable characteristics, and determining an optimal decision variable value meeting a real-time load demand; and S4, making a regulation and control strategy according to the optimal decision variable value, and realizing optimal operation of the micro-grid. According to the method, the load demand of the power system is predicted based on the trained recurrent neural network model, the most critical characteristic variable is identified, the time sequence data which can most represent the power grid load operation state during the microgrid operation period is extracted and analyzed, and the unnecessary calculation amount is reduced by focusing on the most critical characteristic variable, so that the power grid load operation state is optimized. Therefore, the whole analysis process is more efficient and accurate.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of new power system optimization control, in particular to a self-adaptive source-load-storage coordination optimization control method and system. BACKGROUND

[0002] In the process of new power system construction, the micro-grid "source-grid-load-storage" has a unified operation subject. That is, the micro-grid realizes the inter-grid power exchange with the grid in the form of a whole (independent operation subject), and optimizes the internal "source-grid-load-storage" by using energy storage facilities and automatic control systems, to realize self-control and off-grid autonomous management. At present, the integration of source-grid-load-storage optimizes and integrates local power supply side, grid side and load side resources, supported by advanced technologies such as energy storage and institutional mechanism innovation, with the goal of safety, green and efficiency, to innovate power production and consumption mode, and realize deep collaboration of source, grid, load and storage. However, in the planning and design level, the traditional method stays in the basic functions of data statistical analysis, voltage division and power balance, and 110-35 kilovolt capacity ratio checking, without considering the actual meaning of realizing the collaboration between source-grid-load-storage. The disadvantages are reflected in: A single control mode cannot cope with all micro-grid operation conditions, especially in the case of unstable or missing data connection, static or unchanging control strategy model cannot adapt to long-term and seasonal load changes. At the same time, when considering the collaboration between source-grid-load-storage for modeling and predicting load changes, there are a large number of variables that contribute less to model prediction, especially when the micro-grid is offline or demand surges, which causes data complexity, exponentially increases the computational burden of subsequent analysis and model training, and directly leads to the consequence that the power system cannot maintain stability under various operating conditions, and there is a problem of system collapse caused by load fluctuation or insufficient supply. SUMMARY

[0003] In view of the deficiencies in the prior art, the present application aims to provide a self-adaptive source-load-storage coordination optimization control method and system. First, a recurrent neural network model is trained to predict the load demand of the power system, analyze the influence of each feature on the prediction result, identify the most critical features, and reduce and exclude the influence variables that contribute less to model prediction, to improve the model prediction power. Then, by adding the source-load-storage supply and demand balance and the minimum operation cost to construct the objective function, the micro-grid comprehensive dispatching strategy is carried out to avoid system collapse caused by load fluctuation or insufficient supply, and quickly respond to sudden events in the grid, such as offline or demand surge, thereby reducing the impact on system stability. The problems raised in the above background technology are solved.

[0004] To achieve the above objectives, the present invention provides an adaptive source-load-storage coordinated optimization control method, comprising the following steps: S1. Collecting multi-variable load data representing the operation of a microgrid during a specific time period, and performing format, completeness, and consistency checks on the multi-variable load data to form a time-series dataset D reflecting the load changes of the microgrid during the specific time period; S2. Inputting the acquired time-series dataset D into a recurrent neural network model for training and prediction, identifying the key variable features that have the greatest impact on the microgrid load operation status in the microgrid load demand, thereby limiting the extraction of the time-series data set that best represents the microgrid load operation status during the microgrid operation period; S3. Based on the key variable features, determining a comprehensive economic dispatch objective function aimed at minimizing microgrid operating costs and achieving source-load-storage supply-demand balance, constructing a source-load-storage coordinated control model, and simultaneously solving the source-load-storage coordinated control model by setting key variable parameters to determine the optimal decision variable values ​​for meeting real-time load demand in online or offline modes of the microgrid; S4. Formulating control strategies to maintain supply-demand balance in online or offline modes of the microgrid according to the optimal decision variable values, thereby achieving optimal microgrid operation.

[0005] As a second aspect of the present invention, an adaptive source-load-storage coordinated optimization control system is proposed, comprising: a data acquisition module, which collects multi-variable load data representing the operation of a microgrid during a specific time period, and performs format, completeness, and consistency checks on the multi-variable load data to form a time-series dataset D reflecting the load changes of the microgrid during the specific time period; a microgrid load feature identification and state analysis module, which inputs the acquired time-series dataset D into a recurrent neural network model for training and prediction, identifies the key variable features that have the greatest impact on the microgrid load operation state in the microgrid load demand, thereby limiting the extraction of the time-series data set that best represents the microgrid load operation state during the microgrid operation period; a microgrid coordinated control optimization module, which determines a comprehensive economic dispatch objective function based on the key variable features, aiming to minimize the microgrid operating cost and achieve a source-load-storage supply-demand balance, constructs a source-load-storage coordinated control model, and solves the source-load-storage coordinated control model by setting key variable parameters to determine the optimal decision variable values ​​for meeting real-time load demand in the online or offline mode of the microgrid; and a decision module, which formulates control strategies to maintain supply-demand balance in the online or offline mode of the microgrid according to the optimal decision variable values, thereby achieving optimal microgrid operation.

[0006] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. To address the problem that existing models for predicting load changes considering the coordination between power generation, grid, load, and storage suffer from a large number of variables that contribute little to the model's prediction, leading to increased computational burden and hindering overall optimization, this invention proposes a recurrent neural network model for predicting power system load demand. It analyzes the impact of various features on the prediction results, identifies the most critical feature variables, determines the impact of each data point in the feature dataset on the model's prediction performance over time, and extracts and analyzes which time series data best represent the microgrid's load operation status during microgrid operation. This improves data analysis efficiency and reduces unnecessary computation by focusing on the most critical feature variables, making the entire analysis process more efficient and accurate. Once the key feature variables are identified, relevant time series data can be selected from historical data, ensuring that the focus of analysis remains on the variables that best represent microgrid load changes. By focusing on key features, simplifying the dataset, and excluding variables that contribute little to the model's prediction, this approach reduces data complexity and lowers the computational burden of subsequent analysis and model training. 2. To address the problem that existing technologies struggle to guarantee the economic benefits and supply-demand balance of power systems without a unified optimization model, and cannot ensure the overall optimization of microgrid generation, grid, load, and storage through models, this invention proposes a comprehensive scheduling objective function based on load management. This function aims to minimize microgrid operating costs and achieve a balance between source, load, and storage supply and demand. It ensures the power system remains stable under various operating conditions, preventing system collapse due to load fluctuations or insufficient supply. Furthermore, through load management, even when energy storage devices are offline, the remaining power resources can be more effectively dispatched to meet the system's real-time load demands, thus achieving overall optimization of microgrid generation, grid, load, and storage. Attached Figure Description

[0007] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein: Figure 1 This is a flowchart illustrating the adaptive source-load-storage coordinated optimization control method proposed in one embodiment of the present invention. Detailed Implementation

[0008] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0009] The present invention will be further described in detail below with reference to the accompanying drawings, but this is not intended to limit the scope of the invention.

[0010] As a first aspect of the present invention, such as Figure 1 As shown, as an embodiment of the present invention, an adaptive source-load-storage coordinated optimization control method is proposed, comprising the following steps: S1. Collect multi-variable load data representing the operation of the microgrid within a specific time period, and after checking the format, completeness and consistency of the multi-variable load data, form a time series dataset D reflecting the load changes of the microgrid within a specific time period.

[0011] Based on the above technical concept, it should be noted that in step S1, the acquired multi-source load data includes at least price information P from the market operation database to characterize the economic benefits of the microgrid. pri The real-time output data of grid power sources from each power plant in the microgrid operation system, which characterizes the power generation capacity of the microgrid. sou Grid operation data G, obtained from the microgrid operating system, is an important indicator characterizing the stability and security of the microgrid. pri Original microgrid load demand L loa Status data of energy storage devices obtained in the microgrid operation system S sto In practice, price information P pri Based on the market supply and demand ratio D (t) Fuel cost C f(t） And the electricity price P in the previous period prev(t） The weighted sum is obtained from the calculation, while the market supply and demand ratio D (t） Seasonal demand variations, levels of economic activity, and the availability of renewable energy sources need to be considered. Grid power output data O sou Based on the power generation capacity E of each power source i(t） With efficiency η i(t） The sum of the products is used to calculate the power generation capacity and efficiency, which are affected by equipment condition, fuel supply, and environmental conditions. Grid operation data G pri Based on the load L of each line in the microgrid j(t） With line state coefficient S j(t） The product of these factors is used to calculate the line state coefficient S. j(t） This reflects the line's operating status and potential fault conditions. Original microgrid load demand L loa Based on basic load demand L base(t） With load change ΔL (t） The sum and calculation are obtained, where the load change ΔL (t） The impact of special events and weather forecasts on demand needs to be considered. Energy storage device status data S stoBased on the total capacity S of the energy storage system in the microgrid operation system max With used capacity S used(t） The difference is calculated, and it is understandable that in the specific implementation step S1, in order to improve the accuracy and real-time performance of the data, it is necessary to increase the frequency of data collection. At the same time, by adding a detection and processing mechanism for abnormal data, a real-time monitoring and alarm mechanism for the data can be realized.

[0012] S2. Input the acquired time series dataset D into the recurrent neural network model for training and prediction, identify the key variable features that have the greatest impact on the microgrid load operation status in the original microgrid load demand, and thus limit the extraction of the time series data set that best represents the microgrid load operation status during the operation of the microgrid.

[0013] Based on the above technical concept, it should be noted that step S2, in its specific implementation, performs the following operations: S2-1. Normalize each type of multivariate load data in the acquired time series dataset D to ensure that they are on the same scale. Then, construct the time series dataset D from the normalized multivariate load data for training. S2-2, Construct the LSTM model and define the model architecture, where... For the input layer, the number of neurons in the custom input layer is equal to the number of features; in actual calculation, the number of features is taken as 5, that is, the features are price information P from the market operation database. pri Real-time grid power output data from each power plant under the microgrid operation system. sou Grid operation data G obtained from the microgrid operating system pri Original microgrid load demand L loa Status data of energy storage devices obtained in the microgrid operation system S sto ; For LSTM layers, the initial number of LSTM layers is 2 and the number of units per layer is 50. For each LSTM layer, its state is updated using the following formula: In the formula, σ represents the sigmoid activation function, * represents element-wise multiplication, W and b represent the weights and bias parameters, and f t i t O t These are represented as the forget gate, input gate, and output gate, respectively, in C. t Represented as cell state, h t Represented as hidden state, X norm (t) represents the normalized input feature vector; For the output layer, the number of neurons in a custom output layer is equal to the number of predicted targets; S2-3. Based on the trained LSTM model, predict the load demand of the microgrid, analyze the impact of various features on the prediction results, and identify key variable features. It is understandable that when considering the coordination between power source, grid, load, and storage in modeling and predicting load changes, a large number of variables with small contributions to the model prediction increase the computational burden and make overall optimization difficult. This invention proposes to predict the load demand of the power system based on a trained recurrent neural network model, analyze the impact of various features on the prediction results, identify the most critical feature variables, determine the impact of each data point in the feature dataset on the model's prediction performance in the time series, and extract and analyze which time series data best represent the grid load operation status during microgrid operation, thereby improving the efficiency of data analysis. Therefore, the specific implementation steps for identifying the key variable features mentioned above include: S2-31. Prepare and collect price information P during the operation of the microgrid. pri , Mains power output data O sou , power grid operation data G pri Original microgrid load demand L loa Energy storage device status data S sto Multivariate data, including features, are used to form a test dataset. This multivariate data is then cleaned and preprocessed to ensure data quality and consistency. It should be noted that the collected dataset should contain sufficient historical data to reflect the operating status of the microgrid under different conditions. S2-32, Feature Importance Analysis: This method analyzes the impact of each feature on the model's predictive ability by removing individual features. In practice, each feature is removed one by one, and all values ​​of that feature are replaced with the average value of the time series to simulate the situation where the feature is missing. The adjusted test dataset is then used to retrain and predict the LSTM model. After the prediction is completed, the average relative error of the LSTM model's prediction is calculated to evaluate the performance change of the LSTM model after removing specific features. S2-33. For each feature in the retrained and predicted output of the LSTM model, calculate its presence and removal impact on the LSTM model performance, and assign a score S. fi ; S2-34, Based on the score S fiBy ranking all features, the key features with the greatest impact on LSTM prediction performance are identified. Based on these identified key variable features, time series data that best represent the grid load operation during grid operation can be extracted and analyzed. This method not only improves the efficiency of data analysis but also reduces unnecessary computation by focusing on the most critical variables, making the entire analysis process more efficient and accurate. Once the key features are determined, the next step is to filter time series data related to these features from historical data, ensuring that the focus of the analysis remains on the variables that best represent changes in grid load. By focusing on key features, the dataset can be simplified, eliminating variables that contribute little to model prediction and reducing the computational burden of subsequent analysis and model training. Understandably, by performing steps S2-3, the system can identify and evaluate key variable features in microgrid operation, providing a scientific basis for microgrid optimization control and decision-making, helping to improve the operating efficiency and economy of the microgrid, while ensuring the stability and reliability of power supply.

[0014] Because existing technologies lack a unified optimization model, the economic benefits and supply-demand balance of power systems are difficult to guarantee, and overall optimization cannot be ensured through a model. Therefore, this invention also proposes: S3. Based on the characteristics of key variables, determine a comprehensive economic dispatch objective function aimed at minimizing the operating cost of the microgrid and balancing the supply and demand of the source, load, and storage. Construct a source-load-storage coordinated control model. At the same time, by setting key variable parameters, construct constraints and solve the source-load-storage coordinated control model to determine the optimal decision variable values ​​for meeting real-time load demand in the online or offline mode of the microgrid.

[0015] In one embodiment of the present invention, when constructing the source-load-storage collaborative control model in step S3, the following operations are performed: S3-1. Construct a function with the objective of minimizing the all-time operating cost of the microgrid: In the formula, Z represents the total operating cost of the microgrid during the considered time period T, and T represents the total number of time periods for optimization during the considered microgrid operation period. , , , as well as These are respectively represented as the price information P at time t. pri , Mains power output data O sou , power grid operation data G pri Microgrid load demand L loa Energy storage device status data S sto The determined cost function; S3-2. Define key variable parameters to construct constraints, wherein the constructed constraints shall include at least the following: Supply and demand balance constraints ensure that at any point t, the total power generation in the microgrid minus the microgrid load demand plus the net charge and discharge of the energy storage devices in the microgrid operating system must equal zero in order to maintain the supply and demand balance of the system. Grid power supply operation constraints ensure that the operating status of grid power supply is dynamically adjusted according to the power generation and microgrid load demand at time point t in order to meet the real-time changing power demand. Energy storage device state constraints ensure that at each time point t, the energy storage device's state S is... t+1 It equals the state S at the previous time point t plus the charging power of the energy storage device. and discharge power The difference is multiplied by the charging efficiency n and the time interval Δt; Source-load-storage coordinated control constraints ensure the adjusted reference microgrid load demand D ref With the target microgrid load demand L ref and actual microgrid load demand L t and L used to adjust the target microgrid load demand ref (That is, the ideal load value predicted by the recurrent neural network) and the actual microgrid load demand L t (The actual load value monitored at the current moment is directly inherited from the original microgrid load demand L) loa The control gain K differs between the timing instantiation values. dem The reference threshold S and the reference threshold satisfy the following adjustment relationship: .

[0016] Understandably, by constructing an effective source-load-storage coordinated control model, the operating costs of microgrids can be optimized while ensuring system stability and reliability.

[0017] Meanwhile, in step S3, when solving the source-load-storage coordinated control model, it is necessary to use integer programming methods to identify the integer decision variables of nonlinear elements or dynamic constraints existing in the source-load-storage coordinated control model, and perform the following operations: S3-3. Use the simplex method to solve the source-load-storage cooperative control model without integer constraints to obtain the initial optimal solution; S3-4. Assign a unique identifier to each decision variable in the source-load-storage collaborative control model to obtain decision variables with unique numbers; S3-5. Identify and separate the decision variables that need to satisfy integer constraints. After determining the decision variables that need to satisfy integer constraints, separate them from the optimal solution and round them to satisfy the integer constraints. Ensure that the rounded integer value does not exceed its upper limit. It should be noted that the decision variables that satisfy integer constraints include at least the switch status and the equipment capacity, which can only take integer values. S3-6. Based on the rounded integer variable values, update the objective function and all relevant constraints. At the same time, recalculate the objective function value using the new integer variable values ​​to ensure that the rounded solution is still valid. Through this step, it can be ensured that the source-load-storage coordinated control model of the microgrid can still provide an effective optimization solution while satisfying all integer constraints, unlike the defects of existing technologies. This helps to improve the efficiency and reliability of microgrid operation, while ensuring the rational scheduling of all equipment and resources.

[0018] In one embodiment of the present invention, the specific implementation steps for using the simplex method to solve the source-load-storage cooperative control model without integer constraints and obtain the initial optimal solution include: First, all constraints in the microgrid source-load-storage coordinated control model are converted into matrix form, and slack variables or residual variables are introduced to convert inequality constraints into equality constraints. Second, the objective function and constraints are converted into a simplex tableau format that includes at least basic variables, non-basic variables, right-hand side constants, coefficient matrices, and an objective function row. Third, the simplex method is applied iteratively. During iteration, the selection of basic variables is based on choosing a negative test number from the non-basic variables that maximizes the improvement of the objective function value. The selection of leaving basic variables follows the minimum ratio rule as the selection criterion to ensure the feasibility of the solution. The simplex tableau is updated using Gaussian elimination to transform non-basic variables into basic variables, while simultaneously converting the selected basic variables into... The variables are transformed into non-basic variables. Then, after each iteration, the optimality of the solution is checked by determining whether the test numbers of all non-basic variables are not less than zero. If so, the current solution is output as the optimal solution; otherwise, the iteration continues until the test numbers of all non-basic variables are non-negative, at which point the optimal solution is reached. Finally, the decision variable values ​​of the optimal solution are extracted from the final simplex tableau. It should be noted that these values ​​represent the optimal scheduling strategies for power generation, load, and energy storage devices in the microgrid at different time periods. By using the simplex method, the source-load-storage coordinated control model of the microgrid can be effectively solved, resulting in an optimal scheduling scheme with the lowest cost under the premise of satisfying all operational constraints. This helps the microgrid achieve economical and efficient operation while ensuring the reliability and stability of power supply.

[0019] S4. Based on the optimal decision variable values, formulate control strategies to maintain supply and demand balance in the online or offline modes of the microgrid.

[0020] It is understandable that in step S4 above, after obtaining the optimal decision variable values ​​by solving the source-load-storage collaborative control model, in order to achieve the best operation of the microgrid, the regulation strategy for maintaining supply and demand balance in the offline control mode of the microgrid is as follows: the optimal decision variable values ​​are used as inputs and imported into the computer system. Based on the optimal decision variable values, the feasibility of the system under historical data and predetermined targets is evaluated. The power generation plan or the charging and discharging schedule of the energy storage equipment during the operation of the microgrid is adjusted in real time and then applied to production practice to achieve optimal resource allocation and maximize economic benefits, so as to ensure that the microgrid can meet the load demand in different time periods while reducing operating costs.

[0021] When modeling and predicting load changes considering the coordination between power generation, grid, load, and energy storage, a large number of variables contribute little to the model's predictions. These minor variables are particularly noticeable when the microgrid is offline due to faults or when demand surges, leading to data complexity. Therefore, in this embodiment of the invention, a control strategy for maintaining supply and demand balance under the microgrid online control mode is proposed as follows: Real-time data is collected to continuously monitor the instantaneous values ​​of key variable parameters. The collected real-time data is compared with the calculated optimal decision variable values ​​to assess the deviation between the current microgrid operating state and the optimal operating state. Real-time decisions are made to calculate the control parameters that need to be adjusted promptly. Finally, based on the results of the real-time decisions, a control signal is sent from the computer system to the grid power generators or energy storage devices to maintain supply and demand balance and system stability.

[0022] As a preferred embodiment of the present invention, based on the technical concept proposed in the above steps S1-S3, it should be noted that the core of source-load-storage coordination is the ability of energy storage equipment to discharge when power supply is insufficient and charge when power supply is excessive, but its performance will fluctuate with the operating state (such as high temperature causing efficiency to decrease and aging affecting capacity).

[0023] Therefore, when constructing a source-load-storage coordinated control model, it is necessary to dynamically evaluate the real-time operating status of energy storage devices. This is a prerequisite for minimizing the operating cost of microgrids and achieving a balance between source, load, and storage supply and demand.

[0024] Specifically, this involves acquiring multivariate load data in step S1, and quantifying the parameters of these energy storage devices under real-time operating conditions (energy storage device status data S) before constructing the objective function. sto Otherwise, optimizing to "minimize operating costs" would lose its practical meaning (the minimization of microgrid operating costs is highly dependent on the charging and discharging efficiency and economics of energy storage devices). For example, if the state of charge (SOC) (the percentage of the energy storage device's current remaining capacity) is below 20%, forcibly discharging it will damage the device and increase long-term costs; or if the temperature effect is not considered, the actual charging and discharging efficiency will be lower than the model's expectations, leading to a supply-demand imbalance.

[0025] Based on this, the specific execution process of parameters under the real-time operating status of the energy storage device is as follows: The first step is to collect real-time operational data from microgrids and energy storage devices to build a data foundation, including at least: Basic parameters of energy storage equipment: Record static parameters such as the unique identification code, physical installation location, rated capacity, and design charge / discharge efficiency of the energy storage equipment; Dynamic operation data of energy storage: Real-time acquisition of the current SOC and charge / discharge power of the energy storage equipment. , Cell temperature Cumulative number of loops and voltage ripple Dynamic data is used to assess the real-time operating status of energy storage devices.

[0026] The second step is to calculate the data dimension matching indicators. The matching indicators should include at least the following: Energy storage capacity utilization efficiency indicators This is used to measure the effective utilization of energy storage capacity resources, and its calculation formula is: In the formula, Where T is the rated capacity of the energy storage, and T is the length of the statistical period. The data collection interval is used to understand that this indicator reflects the actual level of energy storage capacity utilization within the cycle; the higher the value, the more fully the capacity is utilized. Energy storage health degradation trend indicators It is used to assess the health status and degradation rate of energy storage devices, and its calculation formula is: In the formula, This refers to the nominal temperature of the battery cell. To the upper limit of safe temperature, To design the number of loops, To allow voltage ripple; Weights of each influencing factor ( Through experimental testing or data analysis, it can be understood that the higher the index value, the more severe the degradation of energy storage health. Energy storage failure impact index This is used to quantify the impact of faults on the coordinated operation of energy storage and microgrids. The calculation formula is as follows: In the formula, The duration of the a-th fault is... The fault level coefficient (e.g., minor fault S=1, moderate fault S=3, severe fault S=5) and n is the number of faults within the statistical period. It can be understood that in practical applications, this indicator reflects the comprehensive impact of faults and is used to determine the degree of interference of energy storage equipment faults on coordinated control.

[0027] Step 3: Standardize the... As the core input for energy storage device condition assessment, it is passed to the condition classification stage in the fourth step, and simultaneously processed from the grid-related data (price information P) obtained in step S1. pri Microgrid load demand L loa The current SOC of the energy storage device provides a complete data chain for the source-load-storage collaborative control model.

[0028] The fourth step is to classify and map the operating status of energy storage devices. Specific implementation steps include: a. Based on preprocessing Through continuous state hierarchical mapping function The operating status of energy storage devices is mapped to different performance ranges, so as to convert the actual operating parameters of energy storage devices into clear performance levels and ensure that the optimization model fits the actual capabilities of physical devices.

[0029] b. Define the various operating states of the energy storage equipment: When the energy storage capacity utilization efficiency is high ( Low trend of health decline () The impact of the failure is 0. When ), the corresponding mapping set is obtained: This state is marked as a highly efficient collaborative state, indicating that the energy storage device is performing well and can deeply participate in the power regulation of the microgrid, such as undertaking peak load shaving and valley filling tasks for new energy consumption; If the energy storage capacity utilization efficiency is moderate ( ), and the trend of health decline is controllable. ), low failure impact ( ), to obtain the corresponding mapping set: This state is marked as a stable and coordinated state, indicating that the energy storage equipment is operating stably but health degradation needs to be monitored. In the coordinated control, conventional power regulation is the main method, such as smoothing small load fluctuations and tracking the output fluctuations of new energy sources. Deep charging and discharging should be avoided to prevent accelerated aging. When the energy storage capacity utilization efficiency is low ( ), and a clear trend of health decline ( ), moderate impact of failure ( ), to obtain the corresponding mapping set: This state is marked as the attenuation coordination state, which means that the performance of the energy storage device is significantly degraded at this time. In the coordination control, it only undertakes basic power regulation (such as ensuring a small power compensation within its own charging and discharging safety range), and prioritizes maintaining the basic operation of the equipment to delay the deterioration of the fault. If the energy storage capacity utilization efficiency is extremely low ( ), and a serious trend of health decline ( ), high failure impact ( ), to obtain the corresponding mapping set: This state is marked as a fault-limited state, indicating that the energy storage device has a significant impact when it fails. Its participation in the coordinated control needs to be limited, such as only being activated temporarily in case of emergency power shortage, and triggering the equipment maintenance or replacement process to ensure the overall safety of the microgrid.

[0030] Step 5: Assess the operational status of the categorized energy storage devices. As a characterizing energy storage device status data S sto Key input data for features.

[0031] As a second aspect of the invention, an adaptive source-load-storage coordinated optimization control system is proposed, which operates the optimization control method proposed in steps S1-S4 above. This system includes a data acquisition module that collects multi-variable load data representing the microgrid's operation during a specific time period, and performs format, completeness, and consistency checks on the multi-variable load data to form a time-series dataset D reflecting the load changes of the microgrid during the specific time period. A microgrid load characteristic identification and state analysis module inputs the acquired time-series dataset D into a recurrent neural network model for training and prediction, identifying the key components that have the greatest impact on the microgrid load operation state within the microgrid load demand. The system identifies key variable characteristics to limit the extraction of time-series data sets that best represent the load operation status of the microgrid during its operation. The microgrid collaborative control optimization module determines a comprehensive economic dispatch objective function based on key variable characteristics, aiming to minimize microgrid operating costs and achieve a balance between source, load, and storage supply and demand. It then constructs a source-load-storage collaborative control model. Simultaneously, by setting key variable parameters and establishing constraints, it solves the source-load-storage collaborative control model to determine the optimal decision variable values ​​for meeting real-time load demand in online or offline modes. The decision module, based on the optimal decision variable values, formulates control strategies to maintain supply and demand balance in both online and offline modes, achieving optimal microgrid operation.

[0032] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. An adaptive source-load-storage coordinated optimization control method, characterized in that: Includes the following steps: S1. Collect multi-variable load data representing the operation of the microgrid within a specific time period, and after checking the format, completeness and consistency of the multi-variable load data, form a time series dataset D reflecting the load changes of the microgrid within a specific time period. S2. Input the acquired time series dataset D into the recurrent neural network model for training and prediction, identify the key variable features that have the greatest impact on the microgrid load operation status in the microgrid load demand, and thus limit the extraction of the time series data group that best represents the microgrid load operation status during the operation of the microgrid. S3. Based on the characteristics of the key variables, determine a comprehensive economic dispatch objective function with the aim of minimizing the microgrid operating cost and balancing the supply and demand of the source, load and storage, and construct a source-load-storage coordinated control model. At the same time, by setting key variable parameters, construct constraint conditions and solve the source-load-storage coordinated control model to determine the optimal decision variable values ​​for meeting real-time load demand in the online or offline mode of the microgrid. S4. Based on the optimal decision variable values, formulate control strategies to maintain supply and demand balance in the online or offline modes of the microgrid, so as to achieve optimal operation of the microgrid.

2. The adaptive source-load-storage coordinated optimization control method according to claim 1, characterized in that: In step S1, the acquired multi-source load data includes at least price information P from the market operation database to characterize the economic benefits of the microgrid. pri The real-time output data of grid power sources from each power plant in the microgrid operation system, which characterizes the power generation capacity of the microgrid. sou Grid operation data G, obtained from the microgrid operating system, is an important indicator characterizing the stability and security of the microgrid. pri Microgrid load demand L loa Status data of energy storage devices obtained in the microgrid operation system S sto ,in, The price information P pri Based on the market supply and demand ratio D (t) Fuel cost C f(t） And the electricity price P in the previous period prev(t） The weighted sum is obtained; the grid power supply output data O is obtained through calculation. sou Based on the power generation capacity E of each power source i(t） With efficiency η i(t） The sum of the products is calculated; the power grid operation data G is obtained. pri Based on the load L of each line in the microgrid j(t） With line state coefficient S j(t） The sum of the products is obtained; the microgrid load demand L is calculated. loa Based on basic load demand L base(t） With load change ΔL (t） The energy storage device status data S is obtained through calculation and summation. sto Based on the total capacity S of the energy storage system in the microgrid operation system max With used capacity S used(t） The difference is calculated.

3. The adaptive source-load-storage coordinated optimization control method according to claim 1, characterized in that: In step S2, the following operations are performed: S2-1. Normalize each type of multivariate load data in the acquired time series dataset D to ensure that they are on the same scale. Then, construct the time series dataset D from the normalized multivariate load data for training. S2-2, Construct the LSTM model and define the model architecture, where... For the input layer, the number of neurons in the input layer is defined to be equal to the number of features; For each LSTM layer, the initial number of LSTM layers is 2 and the number of units per layer is 50. Furthermore, for each LSTM layer, its state is updated using the following formula: In the formula, σ represents the sigmoid activation function, * represents element-wise multiplication, W and b represent the weights and bias parameters, and f t i t O t These are represented as the forget gate, input gate, and output gate, respectively, in C. t Represented as cell state, h t Represented as hidden state, X norm (t) represents the normalized input feature vector; For the output layer, the number of neurons in the custom output layer is equal to the number of predicted targets; S2-3. Based on the trained LSTM model, predict the load demand of the microgrid, analyze the impact of various features on the prediction results, and identify key variable features.

4. The adaptive source-load-storage coordinated optimization control method according to claim 3, characterized in that: The specific implementation steps for identifying the key variable features include: S2-31. Prepare and collect price information P during the operation of the microgrid. pri , Mains power output data O sou , power grid operation data G pri Microgrid load demand L loa Energy storage device status data S sto The multivariate data, including features, is used to form a test dataset, and this multivariate data is cleaned and preprocessed to ensure the quality and consistency of the data. S2-32, Feature Importance Analysis: Each feature is removed one by one, and the case of missing features is simulated by replacing all values ​​of the feature with the average value of the time series. The LSTM model is then retrained and predicted using the adjusted test dataset. After the prediction is completed, the performance change of the LSTM model after removing specific features is evaluated by calculating the average relative error of the LSTM model prediction. S2-33. For each feature in the retrained and predicted output of the LSTM model, calculate its presence and removal impact on the LSTM model performance, and assign a score S. fi ; S2-34, Based on the score S fi By ranking all the features, the key features that have the greatest impact on the prediction performance of LSTM are identified.

5. The adaptive source-load-storage coordinated optimization control method according to claim 1, characterized in that: In step S3, when constructing the source-load-storage collaborative control model, the following operations are performed: S3-1. Constructing the objective function: In the formula, Z represents the total operating cost of the microgrid during the considered time period T, and T represents the total number of time periods for optimization during the considered microgrid operation period. , , , as well as These are respectively represented as the price information P at time t. pri , Mains power output data O sou , power grid operation data G pri Microgrid load demand L loa Energy storage device status data S sto The determined cost function; S3-2. Define key variable parameters to construct constraints, wherein the constructed constraints shall include at least the following: Supply and demand balance constraints ensure that at any point t, the total power generation in the microgrid minus the microgrid load demand plus the net charge and discharge of the energy storage devices in the microgrid operating system must equal zero in order to maintain the supply and demand balance of the system. Grid power supply operation constraints ensure that the operating status of grid power supply is dynamically adjusted according to the power generation and microgrid load demand at time point t in order to meet the real-time changing power demand. Energy storage device state constraints ensure that at each time point t, the energy storage device's state S is... t+1 It equals the state S at the previous time point t plus the difference between the charging power and discharging power of the energy storage device multiplied by the charging efficiency n and the time interval Δt; Source-load-storage coordinated control constraints ensure the adjusted reference microgrid load demand D ref With the target microgrid load demand L ref and actual microgrid load demand L t and L used to adjust the target microgrid load demand ref Compared with the actual microgrid load demand L t The difference in control gain K between them dem The reference threshold S and the reference threshold satisfy the following adjustment relationship: .

6. The adaptive source-load-storage coordinated optimization control method according to claim 1 or 5, characterized in that: In step S3, when solving the source-load-storage coordinated control model, it is necessary to use integer programming methods to identify the integer decision variables of nonlinear elements or dynamic constraints existing in the source-load-storage coordinated control model, and perform the following operations: S3-3. Use the simplex method to solve the source-load-storage cooperative control model without integer constraints to obtain the initial optimal solution; S3-4. Assign a unique identifier to each decision variable in the source-load-storage collaborative control model to obtain decision variables with unique numbers; S3-5. Identify and separate the decision variables that need to satisfy integer constraints. After determining the decision variables that need to satisfy integer constraints, separate them from the optimal solution, round them to satisfy the integer constraints, and ensure that the rounded integer values ​​do not exceed their upper limit. S3-6. Based on the rounded integer variable values, update the objective function and all relevant constraints. At the same time, use the new integer variable values ​​to recalculate the objective function value to ensure that the rounded solution is still valid.

7. The adaptive source-load-storage coordinated optimization control method according to claim 6, characterized in that: S3-3. The specific implementation steps for solving the source-load-storage cooperative control model without integer constraints using the simplex method to obtain the initial optimal solution include: First, all constraints in the microgrid source-load-storage coordinated control model are converted into matrix form, and slack variables or residual variables are introduced to convert inequality constraints into equality constraints. Secondly, the objective function and constraints are converted into a simplex tableau format that includes at least basic variables, non-basic variables, right-hand side constants, coefficient matrix, and rows of objective function. Furthermore, the simplex method is applied iteratively. During the iteration, the selection of basic variables is based on selecting a negative test number from the non-basic variables that maximizes the improvement of the objective function value. In the selection of basic variables, the minimum ratio rule is followed as the selection criterion to ensure the feasibility of the solution. The simplex tableau is updated by Gaussian elimination to transform non-basic variables into basic variables, while simultaneously transforming the selected basic variables into non-basic variables. Therefore, after each iteration, the optimality of the solution is checked by determining whether the test numbers of all non-basic variables are not less than zero. If so, the current solution is output as the optimal solution; otherwise, the iteration is repeated until the test numbers of all non-basic variables are non-negative, at which point the optimal solution is reached. Finally, the decision variable values ​​for the optimal solution are extracted from the final simplex tableau.

8. The adaptive source-load-storage coordinated optimization control method according to claim 1, characterized in that: In step S4, after obtaining the optimal decision variable values ​​by solving the source-load-storage collaborative control model, the control strategy for maintaining supply and demand balance in the microgrid offline mode is as follows: the optimal decision variable values ​​are used as inputs and imported into the computer system. Based on the optimal decision variable values, the feasibility of the system under historical data and predetermined targets is evaluated. The power generation plan or the charging and discharging schedule of the energy storage equipment during the microgrid operation is adjusted in real time and then applied to production practice to achieve optimal resource allocation and maximize economic benefits. In the online mode of the microgrid, the established control strategy for maintaining supply and demand balance is as follows: continuously monitor the instantaneous values ​​of key variable parameters in the microgrid by collecting real-time data, compare the collected real-time data with the optimal decision variable values ​​obtained by solving, evaluate the deviation between the current microgrid operating state and the optimal operating state, make real-time decisions, calculate the control parameters that need to be adjusted in a timely manner, and finally, based on the results of the real-time decisions, send control signals to the grid power generators or energy storage devices through the computer system to maintain supply and demand balance and system stability.

9. The adaptive source-load-storage coordinated optimization control method according to claim 2, characterized in that: The process of obtaining multi-source load data in step S1, and before constructing the source-load-storage coordinated control model, requires quantifying the real-time operating status of the energy storage equipment to obtain the energy storage equipment status data S. sto The process is as follows: First, based on the microgrid and energy storage devices, the basic static parameters and dynamic operating data under their real-time operating status are obtained; Secondly, multi-dimensional matching index calculations are performed, including: energy storage capacity utilization efficiency indexes used to measure the effective utilization of energy storage equipment capacity resources. ; Energy storage health degradation trend index used to assess the health status and degradation rate of energy storage devices. Energy storage fault impact index, used to quantify the impact of faults on the coordinated operation of energy storage and microgrids. ; Secondly, after standardizing the indicator data, it is compared with price information P. pri Microgrid load demand L loa Including grid-related data such as the current SOC of the energy storage device, these serve as the core inputs for the energy storage device's condition assessment, and are processed through a continuous state hierarchical mapping function. This involves mapping the operating status of energy storage devices to different performance ranges, thereby converting the actual operating parameters of energy storage devices into specific performance levels. When the energy storage capacity utilization efficiency is high, the health degradation trend is low, and the failure impact is 0, the corresponding mapping set is obtained. Mark this state as a high-efficiency collaborative state and execute peak shaving and valley filling; If the energy storage capacity utilization efficiency is moderate, the health degradation trend is controllable, and the failure impact is low, the corresponding mapping set is obtained. This state is marked as a stable cooperative state, and normal power regulation is performed with limited depth of charge and discharge; When energy storage capacity utilization efficiency is low, health degradation trend is obvious, and fault impact is moderate, the corresponding mapping set is obtained. Mark this state as the attenuation coordination state and perform micro-power compensation; If the energy storage capacity utilization efficiency is extremely low, the health degradation trend is severe, and the failure impact is high, then the corresponding mapping set is obtained. This state is marked as a fault-limited state, triggering equipment maintenance; Finally, the classification status is output as the energy storage device status data S. sto eigenvalues.

10. An adaptive source-load-storage coordinated optimization control system, operating the optimization control method described in claims 1 to 9, characterized in that: include: The data acquisition module collects multi-variable load data representing the operation of the microgrid within a specific time period, and performs format, completeness and consistency checks on the multi-variable load data to form a time series dataset D reflecting the load changes of the microgrid within a specific time period. The microgrid load characteristic identification and state analysis module inputs the acquired time series dataset D into the recurrent neural network model for training and prediction, and identifies the key variable characteristics that have the greatest impact on the microgrid load operating state in the microgrid load demand, thereby limiting the extraction of the time series data set that best represents the microgrid load operating state during the operation of the microgrid. The microgrid collaborative control optimization module determines a comprehensive economic dispatch objective function based on the key variable characteristics, aiming to minimize the microgrid operating cost and balance the supply and demand of source, load and storage. It constructs a source-load-storage collaborative control model and solves the source-load-storage collaborative control model by setting key variable parameters and constructing constraints to determine the optimal decision variable values ​​for meeting real-time load demand in the online or offline mode of the microgrid. The decision-making module formulates control strategies to maintain supply and demand balance in the online or offline modes of the microgrid based on the optimal decision variable values, thereby achieving optimal operation of the microgrid.

Images