Anti-interference control method of optical storage DC microgrid hybrid energy storage system

Abstract

The invention provides an anti-interference control method of an optical storage direct current micro-grid hybrid energy storage system, and relates to the technical field of energy storage systems, comprising the following steps: establishing a multi-source data set; constructing a basic node by using the denoised multi-source data set, taking the basic node as a matching center, and establishing a step node set; backtracking clustering is carried out on the step node set, each round of backtracking clustering result and corresponding step node data are used as step features to be transmitted upwards, and transmission iteration is executed; establishing an extraction feature set of the multi-source data set according to a transmission iteration result, performing disturbance prediction, and establishing a disturbance prediction result; and establishing a balance scheme according to a disturbance prediction result, and carrying out anti-disturbance control management on the hybrid energy storage system of the optical storage DC micro-grid. The technical problem that in the prior art, various complex external disturbances suffered in the operation process of the optical storage direct current micro-grid hybrid energy storage system are difficult to detect and recognize quickly, and a targeted anti-interference control strategy is lacked, so that the system is unstable in operation is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage systems, and particularly to a disturbance rejection control method for a hybrid energy storage system of a photovoltaic-storage DC microgrid. Background Art

[0002] A hybrid energy storage system of a photovoltaic-storage DC microgrid is an integrated power system that combines photovoltaic power generation, energy storage devices, and a DC distribution network, and has great application potential and economic benefits. However, it still faces many technical problems in practical applications, which limit the performance and reliability of the system. Specifically, the operating environment of the photovoltaic-storage DC microgrid system is complex and full of uncertainties. Various external environments and internal factors are likely to generate irregular disturbances. These disturbances are not only difficult to accurately predict, but may also have a serious impact on the operating stability of the system. Existing disturbance control methods are difficult to quickly detect and identify complex disturbances, lack targeted disturbance rejection control strategies, often react slowly or inaccurately, and are difficult to respond to emergencies in a timely manner, thus leading to unstable system operation. Summary of the Invention

[0003] This application provides a disturbance rejection control method for a hybrid energy storage system of a photovoltaic-storage DC microgrid, aiming to solve the technical problem that in the prior art, it is difficult to quickly detect and identify various complex external disturbances during the operation of the hybrid energy storage system of a photovoltaic-storage DC microgrid, and there is a lack of targeted disturbance rejection control strategies, resulting in unstable system operation.

[0004] A disturbance rejection control method for a hybrid energy storage system of a photovoltaic-storage DC microgrid disclosed in this application includes: establishing a multi-source data set, which is constructed by performing system monitoring on the hybrid energy storage system of a photovoltaic-storage DC microgrid after establishing a sensor network. The data of the multi-source data set includes voltage data, current data, temperature data, wind speed data, light intensity data, and load power data; performing data denoising processing on the multi-source data set, constructing basic nodes with the denoised multi-source data set, and taking the basic nodes as the matching center to establish a ladder node set, where each ladder node in the ladder node set is sequentially constructed with the previous ladder node as the matching center; performing backtracking clustering on the ladder node set, and taking the result of each round of backtracking clustering and the corresponding ladder node data as ladder features to be transmitted upward, and performing transmission iteration, where the ladder features are screened by a ladder analysis network before being transmitted upward; establishing an extraction feature set of the multi-source data set according to the transmission iteration result, predicting disturbances based on the extraction feature set, and establishing a disturbance prediction result; establishing a balancing scheme according to the disturbance prediction result, and performing disturbance rejection control management on the hybrid energy storage system of a photovoltaic-storage DC microgrid based on the balancing scheme.

[0005] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0006] By establishing a sensor network to monitor and collect various operation data of the hybrid energy storage system of the photovoltaic and energy storage DC microgrid, including voltage, current, temperature, wind speed, light intensity, etc., a multi-source data set is established, and the current operation state of the system can be obtained in real time. These data provide complete basic information for subsequent analysis; the multi-source data set is subjected to data denoising processing, effectively removing the noise in the multi-source data and ensuring the accuracy of the basic nodes. With the basic nodes as the matching center, a stepped node set is constructed, where each layer of nodes is constructed based on the nodes of the previous layer, forming a progressively layered structure. This hierarchical data structure can better manage, analyze, and process complex multi-source data, providing a basis for backtracking clustering and feature transfer in subsequent steps; backtracking clustering and feature upward transfer optimize the detection and recognition process of disturbances by aggregating feature data layer by layer. Through the screening and layer-by-layer transfer of stepped features, the most representative features can be extracted from a large amount of multi-source data, enhancing the response speed and accuracy of the system in the face of different disturbances; an extracted feature set is established according to the transfer iteration results, and the extracted core features are used for disturbance prediction. According to the disturbance prediction results, factors that may affect stability in the future can be detected in a timely manner, and corresponding countermeasures can be formulated in advance based on these prediction results, improving the preventive control ability of the system; a balancing scheme is established based on the disturbance prediction results to ensure that the system can be adjusted through reasonable resource scheduling and energy management when possible disturbances occur. Through the intelligent scheduling of power generation equipment, energy storage equipment, and loads, the best response strategy can be automatically selected, avoiding the delay of human intervention and ensuring the stability of the system in a changing environment.

[0007] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the specific embodiments of this application are hereinafter specifically exemplified. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 FIG. provides a schematic flowchart of an anti-disturbance control method for a hybrid energy storage system of a photovoltaic and energy storage DC microgrid according to an embodiment of this application;

[0009] Figure 2 FIG. provides a schematic flowchart of establishing a stepped data set in an anti-disturbance control method for a hybrid energy storage system of a photovoltaic and energy storage DC microgrid according to an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0010] By providing a disturbance rejection control method for a hybrid energy storage system of a photovoltaic-storage DC microgrid in an embodiment of the present application, the technical problem in the prior art that it is difficult to quickly detect and identify various complex external disturbances during the operation of the hybrid energy storage system of the photovoltaic-storage DC microgrid, and there is a lack of targeted disturbance rejection control strategies, resulting in unstable system operation, is solved.

[0011] After introducing the basic principle of the present application, various non-limiting implementation manners of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification.

[0012] As Figure 1 shown, an embodiment of the present application provides a disturbance rejection control method for a hybrid energy storage system of a photovoltaic-storage DC microgrid, and the method includes:

[0013] Establish a multi-source data set, which is constructed by performing system monitoring on the hybrid energy storage system of the photovoltaic-storage DC microgrid after establishing a sensor network, and the data of the multi-source data set includes voltage data, current data, temperature data, wind speed data, light intensity data, and load power data.

[0014] Establish a sensor network. The sensor network is a distributed system composed of multiple sensors, including voltage sensors, current sensors, temperature sensors, wind speed sensors, etc. These sensors are installed at various key positions of the hybrid energy storage system of the photovoltaic-storage DC microgrid. After the sensor network is started, each sensor continuously collects relevant data to form a multi-source data set.

[0015] The multi-source data set includes voltage data, current data, temperature data, wind speed data, light intensity data, and load power data. Among them, the voltage data includes voltage changes at key nodes such as photovoltaic panels, energy storage devices, and DC buses; the current data includes the current flow conditions of each device, which is used to evaluate the operating status of each load and power generation unit; the temperature data includes the working temperature of the device, the ambient temperature, etc., which is used to evaluate the heat dissipation capacity of the device and the impact of the environment on the system; the wind speed data can help evaluate the change of power generation; the light intensity data is for the photovoltaic power generation unit, and the light intensity directly determines the power generation efficiency of the photovoltaic panel; the load power data includes the power consumption of each load in the system, such as lighting, household appliances, etc.

[0016] Perform data denoising processing on the multi-source data set, construct a basic node with the denoised multi-source data set, and establish a ladder node set with the basic node as the matching center, where each ladder node in the ladder node set is sequentially constructed with the previous ladder node as the matching center.

[0017] During the acquisition process of multi-source datasets, due to various factors such as sensor errors, environmental interference, signal noise, etc., the original data may contain noise, which will affect the accuracy of the data and the subsequent analysis results. Therefore, it is necessary to denoise the data. Denoising methods include filtering methods, moving average methods, wavelet transforms, etc. Through denoising, more accurate key parameters such as voltage, current, temperature, wind speed, etc. can be extracted, improving the reliability of the data.

[0018] Construct basic nodes based on the multi-source dataset after denoising. The basic nodes are the starting points of data analysis and represent the basic units of the system state. Taking the basic nodes as the matching center, the matching center is a core reference point in data analysis, which means that subsequent data analysis and clustering operations are carried out around the basic nodes. The construction of basic nodes determines the starting point of system analysis and serves as the matching benchmark for nodes at all levels in the data processing process.

[0019] Establish a ladder node set. The ladder node set is a hierarchical data structure composed of multiple nodes constructed in hierarchical order. Each ladder node takes the previous ladder node as the matching center and is constructed layer by layer according to the matching degree and feature similarity. Specifically, starting from the basic nodes, search for the data set in the multi-source dataset that best matches the basic nodes to form the first-level ladder nodes. Take the first-level ladder nodes as the new matching center and find the data that best matches them in the remaining multi-source dataset to construct the second-level ladder nodes. Continuously iterate this process until the complete ladder node set is constructed.

[0020] Perform backtracking clustering on the ladder node set, and transfer the results of each round of backtracking clustering and the corresponding ladder node data upward as ladder features, and execute transfer iteration. Among them, the ladder features are screened through a ladder analysis network before being transferred upward.

[0021] Backtracking clustering is a specific type of clustering algorithm. Its core idea is to start clustering layer by layer from the last layer, that is, the farthest ladder nodes. Different from the conventional bottom-up clustering, backtracking clustering gradually collects data and conducts aggregation analysis by backtracking from the end nodes to the upper nodes. In the ladder node set, each round of clustering operation aggregates the data of the current ladder node to form higher-level data features. The result is a set of new aggregated data, representing the data features at the current ladder level. And the clustering results of each round will be combined with the corresponding ladder node data to form new ladder features, which are more refined and can help better understand the state and trend of the system.

[0022] After the backtracking clustering is completed, the generated ladder features are passed upward to higher-level nodes. This upward passing process is similar to information flow, gradually summarizing the underlying details to the upper layer to form the global features of the system. The upward passing is not only a process of information transmission but also a process of gradual refinement of features. The data will be further screened and aggregated at each level to extract the most valuable features. The transfer iteration is executed. The transfer iteration is a repetitive process. As the backtracking clustering progresses, the ladder features are continuously passed between various ladder nodes. Each round of iteration gradually improves the system's analysis ability of the data by updating the data features, making the features converge upward more accurately and effectively.

[0023] Among them, before the ladder features are passed upward, feature screening is performed through a ladder analysis network. The ladder analysis network is a feature extraction and filtering mechanism that excludes unimportant or noise features by analyzing the relative importance of features and only retains the key features that can best reflect the system state. Feature screening methods include principal component analysis, etc.

[0024] An extraction feature set of the multi-source data set is established according to the transfer iteration result, and disturbance prediction is performed based on the extraction feature set to establish a disturbance prediction result.

[0025] According to the transfer iteration result, the ladder features are integrated to obtain an extraction feature set, and disturbance prediction is performed based on the extraction feature set. Disturbance refers to abnormal situations or fluctuations that may occur during the operation of the system, such as load fluctuations, light changes, wind speed changes, voltage and current abnormalities, etc. Disturbance prediction is to predict the possible external or internal disturbances of the system by analyzing the data patterns in the feature set. Exemplarily, predict the future disturbance trend based on historical data, such as using ARIMA models, long short-term memory networks, etc., and generate a disturbance prediction result through disturbance prediction.

[0026] A balancing scheme is established according to the disturbance prediction result, and anti-disturbance control management of the hybrid energy storage system of the photovoltaic-storage DC microgrid is carried out based on the balancing scheme.

[0027] The core of the balancing scheme is to balance possible supply and demand fluctuations and reduce or eliminate the impact of disturbances on the system by adjusting the resource configuration, energy management, and control strategies in the system. Specifically, first clarify the specific sources of disturbances, such as photovoltaic power generation, energy storage devices, or the load side, etc. According to the characteristics of different disturbance sources, formulate corresponding balancing countermeasures, evaluate the current state of each subsystem in the photovoltaic-storage DC microgrid, including the power reserve of energy storage devices, load demand, power generation capacity of power generation units, etc. By evaluating the current state of the system, decide how to allocate resources to cope with the upcoming disturbances, and finally determine the balancing scheme.

[0028] Implement a balancing scheme in the hybrid energy storage system of a photovoltaic-storage DC microgrid to complete dynamic adjustments in multiple aspects such as power generation, energy storage, and load management, achieve disturbance rejection control management. By executing the balancing scheme, it can ensure that the photovoltaic-storage DC microgrid operates stably in the face of various disturbances, improving the system's disturbance rejection ability and operating efficiency.

[0029] Furthermore, as Figure 2 shown, taking the basic node as the matching center to establish a stepped node set further includes:

[0030] Obtain the node importance of the basic node, and configure a first-level stepped screening channel based on the node importance; taking the basic node as the matching center, perform data matching search in the multi-source data set after denoising processing, and perform matching search and screening through the first-level stepped screening channel to establish a first-level stepped data set; reset the first-level stepped data set as the matching center, perform data matching search in the multi-source data set after denoising processing, and perform matching search and screening through the second-level stepped screening channel to establish a second-level stepped data set, where the second-level stepped screening channel is constructed based on the data matching degree in the first-level stepped data set; perform matching search iteration to establish a stepped data set.

[0031] Node importance is an index to measure the influence and key role of the basic node in the whole system. Node importance can be calculated based on multiple factors. Exemplarily, according to the representativeness of data, some basic nodes contain more representative data such as voltage, current, and temperature, which can more accurately reflect the state of the system, so their importance is higher; according to historical relevance, the historical data of the node has a high correlation with system abnormal events or disturbances, and such nodes have higher value in predicting future disturbances; according to information gain, based on the information gain provided by each node, judge whether the node can bring greater predictive value to subsequent data analysis. Calculate comprehensively based on the above multiple factors to obtain the node importance.

[0032] Configure a first-level stepped screening channel based on node importance. The first-level stepped screening channel is a screening mechanism constructed based on node importance. For example, only data with a node importance higher than a certain threshold can pass the screening. The purpose is to perform matching and screening operations through this channel, give priority to selecting basic nodes with higher importance in the first-level screening, and ensure that the most representative nodes can enter the subsequent matching and analysis process.

[0033] With the base node as the matching center, perform data matching search in the multi-source data set after denoising. The data matching search can use distance-based matching, such as using Euclidean distance, Manhattan distance and other similarity measurement methods to find the data point closest to the base node. After performing the data matching search, the matching results are filtered through the configured first-level ladder screening channel. The screening is mainly based on the importance of the node, matching similarity, etc. The goal of the screening is to remove irrelevant or noisy data points, and only retain data points with high matching degree with the base node and strong predictive value. After matching search and screening, a first-level ladder data set is established. This data set contains the filtered feature data related to the base node.

[0034] Taking the data in the first-level ladder data set as the new matching center is equivalent to taking this batch of data as the new basic node for the next round of data matching. Again, in the multi-source data set that has been denoised, around the new matching center, that is, the first-level ladder data set, a data matching search is performed to find other data with a high matching degree with the data in the first-level ladder data set. The second-level ladder screening channel is used to screen the results of the matching search. The second-level ladder screening channel is built based on the data matching degree in the first-level ladder data set, which means that when the data in the first-level ladder data set matches other data, the screening channel determines which data can pass the screening and enter the second-level data set according to the matching degree of the first-level data set. That is, only the data with a matching degree with the data in the first-level ladder data set higher than a certain threshold can pass the screening. After matching search and screening, a second-level ladder data set is generated. This data set contains other data points with a high matching degree with the data in the first-level ladder data set.

[0035] Continue to perform iterative operations of matching search, that is, repeat the previous matching and screening process, take each level of data set as the new matching center in turn, continue to search for matching data in the multi-source data set, and establish a higher-level ladder data set layer by layer. After each round of iteration, a new ladder data set is generated. The node set of each layer is built on the basis of the data set of the previous layer, and more valuable and representative data are extracted layer by layer. Finally, through multiple matching search iterations, a complete ladder data set is constructed. The ladder data set is a multi-level structure. Each layer of the data set contains data that is highly correlated with the data set of the previous layer, and the key features of the system are gradually extracted.

[0036] Furthermore, the step of performing backtracking clustering on the step node set and transferring each round of backtracking clustering results and corresponding step node data upward as step features to perform transfer iterations further includes:

[0037] Locate the farthest ladder dataset in the ladder node set; perform data aggregation within the farthest ladder dataset through the first aggregation strategy, and use the aggregation result as the bottom-layer backtracking clustering result. Then, use the bottom-layer backtracking clustering result and the farthest ladder dataset as ladder features and pass them upward; obtain the total number of ladders, and configure an aggregation strategy change node based on the total number of ladders; perform an iterative aggregation strategy change for transmission according to the aggregation strategy change node and the second aggregation strategy to complete the transmission iteration.

[0038] The farthest ladder dataset is the node set at the bottom layer or the farthest end in the ladder node set. That is, after multiple iterations and matching screenings, it is the dataset at the layer farthest from the base node. From the constructed multi-layer ladder dataset, find the farthest group of node sets in the current hierarchical structure, which is the farthest ladder dataset. The data in this node set has a lower similarity to the base node or is relatively more dispersed in terms of features.

[0039] The first aggregation strategy aims to merge data with dispersed features through a certain strategy to obtain a more representative aggregation result. Exemplarily, perform central value aggregation and use statistical methods to calculate the central tendency in the dataset, such as the mean, median, mode, etc., to represent the overall characteristics of the entire dataset. This can reduce the influence of noise data and extract the central tendency of the data; eliminate far points. For points that are far away or abnormal in the data, by eliminating these data points far from the concentration, the influence of outliers on the aggregation result can be eliminated to improve the aggregation quality of the data and ensure that the result is more representative of the true system characteristics. Applying the first aggregation strategy to the farthest ladder dataset can effectively aggregate dispersed data points and compress them into a more concise feature representation. The obtained aggregation result represents the main trend or characteristics of the data in the entire farthest ladder node set.

[0040] Use the obtained aggregation result as the bottom-layer backtracking clustering result. Bottom-layer backtracking clustering means starting from the bottom layer, that is, the farthest ladder dataset, and gradually clustering upward, integrating data features layer by layer. The bottom-layer backtracking clustering result represents the core features of the farthest ladder dataset. Use the bottom-layer backtracking clustering result and the farthest ladder dataset together as ladder features and gradually pass them upward to the upper-level node. The process of feature upward transmission is similar to summarizing the bottom-layer aggregation result into higher-level ladder nodes. This operation helps the system gradually integrate bottom-layer and high-level features, making the final features more complete and representative.

[0041] The total number of levels is the number of levels in the set of ladder nodes that have been constructed. By analyzing the constructed ladder data set, the number of levels from the base node to the farthest ladder node is counted to obtain the total number of levels. The aggregation strategy change node is the node position where the aggregation strategy is decided to be changed during the passing iteration. Usually, the data distribution and feature complexity may be different at different levels. The data at lower levels is relatively simple, so a more direct aggregation strategy is used; as the level increases, the data complexity increases, and a more refined aggregation strategy needs to be switched to ensure the quality of data integration.

[0042] Based on the total number of levels in the entire ladder hierarchy, decide at which level to adjust the aggregation strategy and configure the aggregation strategy change node. For example, if the total number of levels is 5, then the aggregation strategy can be set to change at the 3rd or 4th level node. The selection of the change node usually depends on the characteristics of the data and the actual requirements of the system.

[0043] The second aggregation strategy can be set to weighted aggregation. Weighted aggregation is an aggregation method that considers the weights of different data points. Different from simple mean or median aggregation, weighted aggregation assigns a weight to each data point, making the more important or representative data have a greater impact on the aggregation result. This strategy can better reflect the feature distribution of the data, especially at high-level nodes. As the aggregation strategy changes at different levels, the final passing iteration process is completed. In each layer iteration, data integration is performed through different aggregation strategies, and the features of the underlying data are gradually passed upward, finally completing the feature integration of the entire multi-level ladder nodes.

[0044] Furthermore, before the ladder features are passed upward, feature screening is performed through a ladder analysis network, and it also includes:

[0045] Perform window system load analysis on the photovoltaic-energy storage DC microgrid hybrid energy storage system, and establish a first ladder impact factor according to the window system load analysis result; perform feature discretization analysis on the ladder features, and establish a second ladder impact factor according to the feature discretization analysis result; obtain the current ladder level, establish a third ladder impact factor according to the ladder level, and initialize the ladder analysis network according to the first ladder impact factor, the second ladder impact factor, and the third ladder impact factor, and perform feature screening with the initialized ladder analysis network.

[0046] Window system load analysis refers to analyzing and evaluating the load of a photovoltaic-storage DC microgrid system within a specific time window. Exemplarily, using the sliding window technique, the load data is divided into multiple time windows, such as every 5 minutes or every hour, and the average load, peak value, fluctuation range, etc. within each window are analyzed. According to the load analysis results, a first-tier impact factor is generated, which reflects the degree of impact of the load on the system operation. During periods with large load fluctuations, the first-tier impact factor is higher because more energy scheduling and energy storage support are required during these periods.

[0047] Feature discretization analysis is to discretize the step features, evaluate the distribution of different features and the differences between features. The discretized features can more intuitively represent the relationships between features. Exemplarily, continuous step features are divided into multiple intervals, and the frequency distribution and dispersion of feature values within each interval are calculated. A feature with a larger dispersion means that its data distribution is more scattered and the differences between different steps are larger. Based on the results of feature discretization analysis, a second-tier impact factor is generated. If certain features show large differences at different step levels, that is, a large dispersion, it indicates that these features may be of high importance for system disturbance prediction or energy scheduling, so their impact factors will be higher.

[0048] Obtain the current step level, which is the step node level where the current data analysis is located. Based on the current step level, a third-tier impact factor is established to measure the impact of this step level on the overall behavior of the system. Different step levels have different importance, so the weights of the impact factors also need to change accordingly. Specifically, for lower step levels, such as levels close to the basic nodes, the data features are relatively simple but are usually highly correlated with the original state or actual operation of the system, so they have higher impact factors; for higher step levels, the data has been aggregated multiple times and the degree of feature abstraction is higher. Although the features may be more complex, their impact on the system may be relatively small, so the impact factors can be set lower.

[0049] Input the first-tier impact factor, the second-tier impact factor, and the third-tier impact factor into the step analysis network as initial weights to guide the network's decision-making in subsequent feature screening, making the feature screening process more efficient and accurate.

[0050] Furthermore, the establishment of the balance plan according to the disturbance prediction result further includes:

[0051] Locate the disturbance source using the disturbance prediction result, evaluate the self-steady-state control based on the disturbance source, and establish a self-optimization evaluation result; obtain the device tasks of the hybrid energy storage system of the photovoltaic-storage DC microgrid, and perform an optimization evaluation according to the device tasks and the disturbance source to establish a replacement optimization evaluation result; perform scheme optimization based on the self-optimization evaluation result and the replacement optimization evaluation result to establish a balance scheme.

[0052] Identify the specific factors or sources that cause the disturbance according to the disturbance prediction result to obtain the disturbance source, which includes load-side disturbance, power generation-side disturbance, energy storage system disturbance, etc. The self-steady-state control evaluation is to evaluate the current operating state of the disturbance source and judge whether it can maintain stability when disturbed. Through the self-steady-state control evaluation, a self-optimization evaluation result is established, that is, to evaluate whether the system can cope with the disturbance through its own control strategy without external intervention. This result reflects the self-adjustment ability of the system and its stability under disturbance conditions.

[0053] The device tasks are the specific functions and objectives of various devices in the photovoltaic-storage DC microgrid system at different time periods. For example, the task of the photovoltaic power generation device is to generate the maximum power according to the light intensity; the task of the energy storage device is to discharge when the power supply is insufficient and charge when the supply is excessive; the task of the load is to operate stably or adjust the power consumption according to the demand.

[0054] The optimization evaluation is to evaluate how each device adjusts its working state to achieve the optimal system stability when the disturbance source is known. For example, if it is predicted that the photovoltaic power generation device will be disturbed, such as a decrease in light intensity, the energy storage device can be adjusted to discharge in advance to make up for the insufficient power generation; if the load suddenly increases, the scheduling strategies of the energy storage device and the power generation device can be adjusted to ensure that there is no power gap. Through the optimization evaluation, a replacement optimization evaluation result is established, that is, when the system cannot maintain a steady state through self-adjustment, that is, self-optimization, an optimization adjustment is made based on the device tasks to replace the original working method to maintain system stability.

[0055] Based on the self-optimization evaluation result and the replacement optimization evaluation result, find the best balance between stability and resource utilization efficiency, determine the optimal balance scheme to cope with the system disturbance. Exemplarily, first, according to the self-optimization evaluation result, judge whether the system can maintain stability by adjusting its own control strategy. If so, this scheme is preferentially adopted. If the self-optimization is insufficient to cope with the disturbance, combine the replacement optimization evaluation result, analyze the scheme to cope with the disturbance by device scheduling or load adjustment, and select a comprehensive optimal scheme according to the results of self-optimization and replacement optimization, which can effectively cope with the disturbance and maximize the system operation efficiency, reduce energy loss or resource waste.

[0056] Furthermore, the method further includes:

[0057] Read the energy storage data of the energy storage device; if both the self-optimization evaluation result and the replacement optimization evaluation result cannot meet the preset evaluation threshold, generate an energy storage compensation instruction; generate an energy storage compensation control plan based on the energy storage compensation instruction and the energy storage data, and use the energy storage compensation control plan as the balancing plan.

[0058] Read the energy storage data of the energy storage device. The energy storage data refers to the current state information of the energy storage device, including battery power, charge and discharge status, available capacity, charge and discharge efficiency, etc.

[0059] The preset evaluation threshold refers to the stability or performance standard set during system operation, which may include the power supply and demand balance threshold, the system voltage and current stability threshold, etc. If both the self-optimization and replacement optimization results are lower than the preset threshold, it means that the system cannot maintain stability through normal energy scheduling and optimization strategies. At this time, generate an energy storage compensation instruction to require the energy storage device to intervene.

[0060] Generate an energy storage compensation control plan based on the energy storage compensation instruction and the energy storage data. Exemplarily, generate a discharge strategy to determine the discharge power, time, and duration of the energy storage device. For example, when the system power supply is insufficient, the energy storage device discharges at the maximum power for a period of time until the system returns to stability; generate a charging strategy. If the energy storage device has a low battery level, it charges during periods of low power demand to ensure that it can handle new disturbances in the future; generate a dynamic scheduling strategy to dynamically adjust the charge and discharge behavior of the energy storage device according to real-time supply and demand changes to ensure that the system always maintains a stable state.

[0061] In the case where the system cannot be stabilized through ordinary scheduling and optimization strategies, use the energy storage compensation control plan as the final balancing plan to ensure that the system can still operate stably when encountering disturbances.

[0062] Furthermore, the method further includes:

[0063] Establish an instantaneous compensation demand based on the disturbance prediction result; establish an additional penalty factor based on the instantaneous compensation demand, and perform a solution optimization penalty on the self-optimization evaluation result and the replacement optimization evaluation result based on the additional penalty factor, and update the balancing plan according to the penalty result.

[0064] Analyze the instantaneous disturbance based on the disturbance prediction result. The instantaneous disturbance includes a sudden increase in load, voltage fluctuation, or abnormal output of the power generation device within a short period of time, etc. Calculate the power required to cope with these instantaneous disturbances. For example, when the load suddenly increases, calculate how much power or electricity the energy storage device needs to provide in a short period of time to make up for the supply-demand gap, and establish an instantaneous compensation demand based on the calculation result.

[0065] An additional penalty factor is established based on the instantaneous compensation requirement to penalize those optimization solutions that cannot quickly meet the compensation requirement. The magnitude of the penalty factor depends on the delay of the compensation response and the degree of insufficient compensation. Exemplarily, if an optimization solution cannot meet the instantaneous compensation requirement within the specified time, a relatively large penalty factor will be assigned. For example, if the system predicts that a sudden increase in load requires additional power to be provided within 1 second, and a certain solution can only respond within 3 seconds, this solution will be severely penalized; if the compensation amount provided by the optimization solution cannot fully cover the instantaneous demand, it will also be penalized. For example, if the load demand increases by 10 kW, but a certain solution can only provide 8 kW of compensation, such a solution will be assigned a relatively high penalty factor.

[0066] Based on the additional penalty factor, scheme optimization penalty is performed to impose penalties on those optimization solutions with untimely responses or insufficient compensation, making them at a disadvantage in scheme selection, and then preferentially selecting those solutions that can respond quickly and have strong compensation capabilities. According to the results after imposing the penalties, the updated balance scheme is determined.

[0067] Furthermore, the method further includes:

[0068] Based on the disturbance prediction results, periodic disturbance data is established, and a decision-making compensation scheme is generated using the periodic disturbance data; the task execution of the hybrid energy storage system of the photovoltaic-storage DC microgrid is optimized through the decision-making compensation scheme.

[0069] Periodic disturbances are disturbances that occur periodically due to certain regular factors, such as time, weather, load demand, etc. Such disturbances are usually predictable because they are closely related to the system's operating mode, external environment, or user behavior. Based on the disturbance prediction results, periodic disturbance data is established. For example, the power generation of photovoltaic power generation will show regular periodic fluctuations with the daily sunlight changes; the electrical load may cycle between the peak and trough of a day.

[0070] Generate a decision compensation plan based on periodic disturbance data, aiming to enable the system to automatically adjust its operating state, perform energy scheduling and load management when periodic disturbances occur. Specifically, through the periodic disturbance data, predict possible load changes or power generation fluctuations during certain periods. For example, during the peak period of photovoltaic power generation, there may be an excess of electricity; during the peak period of night-time electricity consumption, there may be a shortage of electricity. Based on this, formulate compensation strategies. For example, during the period of excess electricity, arrange for energy storage devices to charge and store the excess electrical energy; during the period of insufficient electricity, arrange for energy storage devices to discharge, or advance the scheduling of standby power generation units to meet the load demand; for periods with large periodic changes, it is also possible to perform peak shaving and valley filling in advance, arrange for non-critical loads to operate with a time delay, and reduce the pressure on the system. Considering the capabilities of energy storage devices and power generation devices, finally determine the decision compensation plan to ensure that the supply-demand balance can be maintained during periodic disturbances.

[0071] By implementing the decision compensation plan, optimize various tasks in the hybrid energy storage system of the photovoltaic-storage DC microgrid, such as power generation, energy storage, load management, etc., to maximize the system efficiency and stability.

[0072] In summary, the anti-disturbance control method for a hybrid energy storage system of a photovoltaic-storage DC microgrid provided by the embodiments of this application has the following technical effects:

[0073] By establishing a sensor network to monitor and collect various operation data of the hybrid energy storage system in the photovoltaic and energy storage DC microgrid, including voltage, current, temperature, wind speed, light intensity, etc., a multi-source data set is established, which can obtain the current operation state of the system in real time. These data provide complete basic information for subsequent analysis; the multi-source data set is denoised, effectively removing the noise in the multi-source data and ensuring the accuracy of the basic nodes. With the basic nodes as the matching center, a stepped node set is constructed, where each layer of nodes is constructed based on the nodes of the previous layer, forming a progressive structure. This hierarchical data structure can better manage, analyze and process complex multi-source data, providing a basis for backtracking clustering and feature transfer in the subsequent steps; backtracking clustering and feature upward transfer optimize the detection and recognition process of disturbances by aggregating feature data layer by layer. Through the screening and layer-by-layer transfer of stepped features, the most representative features can be extracted from a large amount of multi-source data, enhancing the response speed and accuracy of the system in the face of different disturbances; an extracted feature set is established according to the transfer iteration results, and the extracted core features are used for disturbance prediction. According to the disturbance prediction results, factors that may affect stability in the future can be detected in a timely manner, and corresponding countermeasures can be formulated in advance based on these prediction results, improving the preventive control ability of the system; a balancing scheme is established based on the disturbance prediction results to ensure that the system can be adjusted through reasonable resource scheduling and energy management when possible disturbances occur. Through the intelligent scheduling of power generation equipment, energy storage equipment and loads, the best response strategy can be automatically selected, avoiding the delay of human intervention and ensuring the stability of the system in a changing environment.

[0074] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An anti-disturbance control method for a photovoltaic-storage-DC microgrid hybrid energy storage system, characterized in that: The method comprises: Establish a multi-source data set, wherein the multi-source data set is constructed by executing system monitoring of a photovoltaic-storage-DC microgrid hybrid energy storage system after establishing a sensor network, and the data of the multi-source data set includes voltage data, current data, temperature data, wind speed data, light intensity data, and load power data; Performing data denoising on the multi-source data set, constructing basic nodes with the denoised multi-source data set, and establishing a ladder node set with the basic nodes as matching centers, wherein each ladder node in the ladder node set is sequentially constructed with the previous ladder node as the matching center; Performing back-tracing clustering on the ladder node set, and passing each round of back-tracing clustering results and corresponding ladder node data upward as ladder features, and performing transfer iterations, wherein the ladder features are subjected to feature screening through a ladder analysis network before being passed upward; Establishing an extraction feature set of a multi-source data set according to the transfer iteration result, performing disturbance prediction based on the extraction feature set, and establishing a disturbance prediction result; A balancing scheme is established according to the disturbance prediction results, and anti-disturbance control management of the photovoltaic-storage-DC microgrid hybrid energy storage system is performed based on the balancing scheme.

2. The anti-disturbance control method of a photovoltaic-storage-DC microgrid hybrid energy storage system according to claim 1, characterized in that: The step of establishing a ladder node set using the basic node as a matching center also includes: Obtaining the node importance of the basic node, and configuring a first-level ladder screening channel based on the node importance; Using the basic node as a matching center, performing data matching search in the multi-source data set after denoising, and performing matching search screening through the first-level ladder screening channel to establish a first-level ladder data set; Resetting the primary ladder data set to a matching center, performing a data matching search in the multi-source data set after denoising, performing a matching search and screening through a secondary ladder screening channel, and establishing a secondary ladder data set, wherein the secondary ladder screening channel is constructed through the data matching degree in the primary ladder data set; Perform matching search iterations to build a ladder dataset.

3. The anti-disturbance control method of a photovoltaic-storage-DC microgrid hybrid energy storage system according to claim 1, characterized in that: The step of performing backtracking clustering on the step node set, and transferring each round of backtracking clustering results and corresponding step node data upward as step features, and performing transfer iterations, further includes: Locate the farthest ladder data set in the ladder node set; Aggregate the data in the farthest step dataset using the first aggregation strategy, and use the aggregation result as the bottom-level backtracking clustering result, and pass the bottom-level backtracking clustering result and the farthest step dataset upward as step features; Get the total number of steps and configure the aggregation strategy change node based on the total number of steps; The aggregation strategy of the transmission iteration is changed according to the aggregation strategy change node and the second aggregation strategy to complete the transmission iteration.

4. The anti-disturbance control method of a photovoltaic-storage-DC microgrid hybrid energy storage system as claimed in claim 3, characterized in that: The step features are screened by a step analysis network before being passed upward, and further include: Performing a window system load analysis on the photovoltaic-storage-DC microgrid hybrid energy storage system, and establishing a first-order impact factor according to the window system load analysis result; Performing feature discrete analysis on the step features, and establishing a second-step impact factor according to the feature discrete analysis results; The current ladder level is obtained, a third ladder impact factor is established according to the ladder level, a ladder analysis network is initialized according to the first ladder impact factor, the second ladder impact factor, and the third ladder impact factor, and feature screening is performed with the initialized ladder analysis network.

5. The anti-disturbance control method of a photovoltaic-storage-DC microgrid hybrid energy storage system according to claim 1, characterized in that: The establishing of a balancing scheme according to the disturbance prediction result also includes: Using the disturbance prediction result to locate the disturbance source, performing self-steady-state control evaluation based on the disturbance source, and establishing a self-optimization evaluation result; Obtaining equipment tasks of the photovoltaic-storage-DC microgrid hybrid energy storage system, performing optimization evaluation according to the equipment tasks and the disturbance sources, and establishing replacement optimization evaluation results; Based on the self-optimization evaluation results and the replacement optimization evaluation results, a solution optimization is performed to establish a balanced solution.

6. The anti-disturbance control method of a photovoltaic-storage-DC microgrid hybrid energy storage system as claimed in claim 5, characterized in that: The method further comprises: Read the energy storage data of the energy storage device; If the self-optimization evaluation result and the replacement optimization evaluation result both fail to meet the preset evaluation threshold, generating an energy storage compensation instruction; An energy storage compensation control scheme is generated according to the energy storage compensation instruction and the energy storage data, and the energy storage compensation control scheme is used as a balancing scheme.

7. The anti-disturbance control method of a photovoltaic-storage-DC microgrid hybrid energy storage system as claimed in claim 5, characterized in that: The method further comprises: Establishing instantaneous compensation requirements according to the disturbance prediction results; An additional penalty factor is established based on the instantaneous compensation demand, and the self-optimization evaluation result and the replacement optimization evaluation result are punished based on the additional penalty factor, and the balance plan is updated according to the penalty result.

8. The anti-disturbance control method for a photovoltaic-storage-DC microgrid hybrid energy storage system according to claim 1, characterized in that: The method further comprises: Establishing periodic disturbance data according to the disturbance prediction result, and generating a decision compensation scheme with the periodic disturbance data; The decision compensation scheme is used to optimize the task execution of the photovoltaic-storage-DC microgrid hybrid energy storage system.

Images