An abnormal state warning method for an energy storage system

Abstract

This application relates to the technical field of energy storage system early warning, and provides an abnormal state early warning method for an energy storage system. The method includes: connecting a monitoring device through an embedded platform interface to obtain operation status data; connecting to a historical record library and extracting a case set based on the status data; analyzing the case set to determine the influence relationship; constructing an abnormal judgment model according to the influence relationship; inputting the monitoring data into the model for abnormal probability analysis and outputting a judgment result; generating a warning message according to the result and loading it into the historical record library to update the abnormal database. This application solves the technical problem that in the existing early warning, due to insufficient data analysis depth methods and low accuracy of the early warning model, the abnormal state of the energy storage system cannot be discovered in a timely and accurate manner, and realizes the technical effect that through the combination of real-time monitoring data and the abnormal judgment model, the abnormal state of the energy storage system can be timely warned, the false negative rate of the early warning can be effectively reduced, and the accuracy and timeliness of the early warning can be improved.

Description

Technical Field

[0001] This application relates to the technical field of energy management, specifically to the technical field of energy storage system early warning, and particularly to an abnormal state early warning method for an energy storage system. Background Art

[0002] With the rapid development and wide application of energy storage technology, energy storage systems play an increasingly important role in power systems. However, due to the complexity and variability of the internal operating state of energy storage systems, the timely detection and handling of their abnormal states have become an urgent problem to be solved. Existing early warning methods for energy storage systems often fail to detect abnormal states in a timely and accurate manner due to insufficient data analysis depth and low accuracy of early warning models, resulting in a high false negative rate of early warnings and posing potential safety hazards to the stable operation of power systems. Summary of the Invention

[0003] This application provides an abnormal state early warning method for an energy storage system, aiming to solve the technical problem that existing early warnings cannot detect the abnormal state of the energy storage system in a timely and accurate manner due to insufficient data analysis depth methods and low accuracy of early warning models.

[0004] In view of the above problems, this application provides an abnormal state early warning method for an energy storage system.

[0005] This application provides an abnormal state early warning method for an energy storage system. The method includes: connecting a monitoring device through an embedded platform interface to obtain operation state data, including power, charge and discharge state, temperature, voltage, and insulation resistance; connecting to a historical record library, and based on the power, charge and discharge state, temperature, voltage, and insulation resistance, extracting an abnormal accident case set, where the abnormal accident case set includes one or more abnormal operation state data among the power, charge and discharge state, temperature, voltage, and insulation resistance; performing data analysis on each operation state of the abnormal accident case set to determine the influence relationship of each operation state data; constructing an abnormal judgment model according to the influence relationship of each operation state data; inputting the operation state data collected by the monitoring device into the abnormal judgment model for abnormal probability analysis, and outputting an abnormal judgment result; generating a warning message according to the abnormal judgment result, and loading the abnormal judgment result into the historical record library to update the abnormal record database.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The above-mentioned abnormal state warning method for an energy storage system connects monitoring devices through an embedded platform interface to obtain key operation state data of the energy storage system in real time, such as power, charge and discharge state, temperature, voltage, insulation resistance, etc. Subsequently, it connects to the historical record library and extracts a case set related to abnormal accidents based on these operation state data. This case set contains data of various abnormal operation states. After that, it deeply analyzes the abnormal accident case set to determine the mutual influence relationship between each operation state data. Based on these influence relationships, an accurate abnormal judgment model is constructed. This model can receive the operation state data collected by the monitoring device in real time, perform abnormal probability analysis, and output the abnormal judgment result. Once an abnormal state is judged, a warning message will be generated immediately to notify the user in a timely manner. At the same time, these abnormal judgment results will also be loaded into the historical record library to continuously update and improve the abnormal record database, and enhance the accuracy and reliability of the warning model. This method realizes the timely warning of the abnormal state of the energy storage system through real-time monitoring and in-depth data analysis, and improves the safety and operation efficiency of the energy storage system.

[0008] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present 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 the present application more obvious and understandable, the specific embodiments of the present application are specifically given below. Description of the Drawings

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0010] Figure 1 It is a schematic flowchart of an abnormal state warning method for an energy storage system in an embodiment;

[0011] Figure 2 It is a schematic diagram of determining the influence relationship of each operation state data of an abnormal state warning method for an energy storage system in an embodiment. Detailed Embodiments

[0012] The embodiments of the present application provide an abnormal state warning method for an energy storage system to solve the technical problem that in the existing warning, due to insufficient depth of data analysis method and low accuracy of the warning model, the abnormal state of the energy storage system cannot be found in time and accurately.

[0013] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0014] It should be noted that the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0015] Embodiment 1

[0016] As Figure 1 shown, the present application provides an abnormal state warning method for an energy storage system, and the method includes:

[0017] Connect to the monitoring device through the embedded platform interface to obtain the operating state data, including power, charge and discharge state, temperature, voltage, and insulation resistance;

[0018] As an energy storage device, the energy storage system can balance power supply and demand and improve the stability of the power system, which is crucial for modern industrial production, power supply and other fields. The embedded platform, on the other hand, is an important part of the energy storage system. It has high integration and real-time performance, and can realize real-time monitoring and data processing of various operating states of the energy storage system. By making full use of the advantages of the embedded platform, connecting the monitoring device, and obtaining the key operating state data of the energy storage system in real time, accurate warning of abnormal states can be realized.

[0019] In the embodiment of the present application, the system terminal successfully connects to the monitoring device through the embedded platform interface, thereby obtaining the key operating state data of the energy storage system in real time. These data cover important information such as power, charge and discharge state, temperature, voltage, and insulation resistance, providing necessary data support for subsequent abnormal state warning. In this way, the operating state of the energy storage system can be comprehensively understood, laying a solid foundation for subsequent analysis and warning work.

[0020] Connect to the historical record library, and based on the power, charge and discharge state, temperature, voltage, and insulation resistance, extract the abnormal accident case set, where the abnormal accident case set includes one or more abnormal operating state data among the power, charge and discharge state, temperature, voltage, and insulation resistance;

[0021] In one embodiment, the system terminal establishes a stable connection with the historical record library by calling the interface of the embedded platform. This step ensures that the platform can smoothly access and retrieve historical data. Subsequently, the system terminal filters the historical record library based on the currently real-time monitored operation state data such as power, charge and discharge state, temperature, voltage, insulation resistance, etc. The purpose of filtering is to find historical records similar to or related to the current state data. Among the filtered historical records, the system terminal circularly retrieves the historical record library according to the currently real-time monitored data. In each retrieval, the system terminal compares the currently real-time monitored data with the retrieval result. If the data is roughly the same, this retrieval is extracted until all the data in the historical record library has been retrieved. These retrieved cases involve various situations such as abnormal power, abnormal charge and discharge state, abnormal temperature, voltage fluctuation, reduction of insulation resistance, etc. Based on the extracted cases, the system terminal will further analyze and organize them to construct an abnormal accident case set. This case set contains various abnormal state data that occurred in the past, which helps to deeply understand various abnormal situations that may occur in the energy storage system. By analyzing these case sets, the correlation and influence relationship between the operation state data can be found, providing strong data support for the subsequent construction of an abnormal judgment model.

[0022] Analyze the operation state data of the abnormal accident case set to determine the influence relationship of each operation state data;

[0023] In one embodiment, restricting too much the analysis of the operation state data of the abnormal accident case set is to deeply explore the correlation and influence relationship between different state data. This process involves deeply analyzing multiple key indicators such as power, charge and discharge state, temperature, voltage, insulation resistance, etc. Through accident type clustering and accident concurrent state analysis, the system terminal can establish the potential connection between them. For example, when the battery temperature rises abnormally, its insulation resistance will decrease accordingly. Through such analysis, the system terminal can not only more comprehensively understand the operation state of the energy storage system, but also provide an important reference basis for subsequent abnormal warning and fault troubleshooting.

[0024] Further, as Figure 2 shown, this application provides a method for analyzing the operation state data of the abnormal accident case set to determine the influence relationship of each operation state data, and the method further includes:

[0025] Perform accident type clustering on the abnormal accident case set to obtain multiple accident sets;

[0026] Respectively perform vertical comparison of the operation state data on the multiple accident sets to obtain the change relationship of the operation state data, and fit the influence relationship between the operation state data and the accident type based on the change relationship of the operation state data;

[0027] Preferably, clustering the accident types of the abnormal accident case set is to classify accidents with similar characteristics and causes into one category, forming multiple different accident sets. Specifically, the system terminal determines the key features for clustering according to the actual situation and historical experience of the accident cases, such as power change, temperature trend, voltage fluctuation, etc. Then, these features are extracted from each case to form a feature vector. Subsequently, the extracted feature vectors are used as input data. These feature vectors contain the key information of each abnormal accident case and are used to describe the similarity between cases. After that, the system terminal randomly selects K data points as the initial cluster centers. Then, for each data point, calculate its distance from each cluster center and assign it to the cluster with the closest distance. Then, for each cluster, recalculate the average value of all data points inside it as the new cluster center. Repeat the above process until the cluster centers no longer change significantly or reach the preset number of iterations. After the clustering operation is completed, the system terminal classifies similar abnormal accident cases into one category according to the output result of the clustering algorithm, forming multiple different accident sets. The cases within each accident set are similar in characteristics and represent the same type of abnormal accident.

[0028] After obtaining multiple accident sets, for each accident set, the system terminal extracts the operation status data of all corresponding abnormal accident cases. Subsequently, the distribution of each operation status data before and after the accident is statistically analyzed, and statistics such as the mean and standard deviation before and after the accident are calculated. These statistics can help the system terminal understand the data distribution and degree of change. Then, the changes in the statistics of the same operation status data in different accident sets before and after the accident are compared, and the statistics with relatively large changes are focused on, as they may have a closer relationship with the accident type. Then, based on the calculated statistics and operation status data, the Pearson correlation coefficient is used to calculate the correlation between operation status data and between them and the accident type, and combined with the degree of change of the statistics, the operation status data significantly related to the accident type is identified. Further, time series analysis is performed on the operation status data significantly related to the accident type to observe the change trend, fluctuation, etc. of the data before and after the accident. Through time series analysis, abnormal patterns of the operation status data before the accident are identified, and these patterns indicate the occurrence of the accident. Subsequently, by comparing and analyzing the data of multiple accident sets, the change patterns of the operation status data related to a specific accident type are identified. These patterns include the continuous change of a single data, the combined change between multiple data, or the occurrence of a specific event sequence. For example, if it is observed that the charging status of the battery is abnormal (such as rapid charging or discharging), and at the same time the battery power drops rapidly, this indicates that there is an internal fault or external load abnormality in the battery. This combined change pattern between the charging status and the battery power can indicate potential accident risks. Then, by comparing the cases in different accident sets, the common points and differences are found, and the associations and differences between different accident types are deeply understood. Through induction and summary, a qualitative description of the influence relationship between the operation status data and the accident type is formed. These descriptions include the thresholds of key indicators, the characteristic differences between different accident types, etc. Through the above analysis process, the relationship between the operation status data and the accident type can be more comprehensively understood, and strong support can be provided for establishing the influence relationship between abnormal events and each operation status data in the follow-up.

[0029] Perform accident concurrency status analysis on the multiple accident sets, determine the concurrent accident types, construct an accident relationship tree, and determine the top event and basic events;

[0030] According to the influence relationship, respectively fit the influence relationship of each operation status data for the top event and the basic events, and establish the influence relationship between abnormal events and each operation status data.

[0031] Preferably, for the analysis of multiple accident sets, the system terminal mainly conducts an analysis of the concurrent states between accidents, that is, the situation where multiple accidents occur simultaneously or successively. Specifically, the system terminal conducts a detailed interpretation of the multiple accident sets, including the occurrence time, location, involved equipment or systems, consequences caused by the accidents, etc. of each accident. Then, a preliminary analysis is carried out on the interpreted data to identify whether there are concurrent accident situations, that is, multiple accidents occur simultaneously or successively. Subsequently, according to the nature, causes, and impacts of the accidents, the concurrent accidents are classified. For example, the accidents are classified into different types such as equipment failures, operation errors, external environmental impacts, etc. The characteristics and laws of each type of concurrent accident are determined to provide a basis for subsequent analysis. After that, taking each accident as a node, they are connected according to the causal relationships between them to construct an accident relationship tree. In the accident relationship tree, starting from the final accident result, the top event is determined. The top event is the end point of the entire accident sequence and also the main consequence of the entire accident. Then, starting from the top event, the cause events that led to the occurrence of this result are traced layer by layer upward to find the direct cause or prerequisite conditions for the occurrence of the top event, that is, the basic events. These basic events are the root causes or starting points of the accident occurrence.

[0032] After determining the top event and the basic events, the system terminal analyzes the influence relationship between the previously determined operating state data and the accident types, clarifies which operating state data have a significant impact on the accident types, and understands the nature and trend of these influence relationships, providing a basis for subsequent analysis of the top event and the basic events. Subsequently, the operating state data related to the top event is identified. These data are closely related to the accident types that led to the top event in the influence relationship between the operating state data and the accident types. Then, the changes in these operating state data when the top event occurs are analyzed, including fluctuations in values, changes in trends, etc., to determine which operating state data have a direct and significant impact on the top event and describe the specific influence relationships between them. For the basic events, the same method is used to identify the related operating state data and analyze the change characteristics of these operating state data when the basic events occur to understand their contribution degrees to the occurrence of the basic events. Then, it is determined which operating state data play a key role in the basic events and describe the specific influence relationships between them. After that, a comprehensive analysis is carried out on the influence relationships between the top event and the basic events and the operating state data to form an overall influence relationship framework. In this framework, the causal relationships and interactions between the abnormal events and each operating state data are clarified, and the specific manifestation forms of these influence relationships are described, thereby establishing the influence relationship between the abnormal events and each operating state data.

[0033] In summary, this process combines multiple steps such as accident concurrent state analysis, accident relationship tree construction, and impact relationship fitting, aiming to comprehensively reveal the complex relationship between operation state data and accident types, providing strong support for preventing and controlling the occurrence of accidents.

[0034] Construct an anomaly judgment model according to the impact relationships of the respective operation state data;

[0035] In one embodiment, the system terminal determines the impact relationships between operation state data and different accident types through analysis. These relationships reflect the internal connection between data changes and accident occurrences and are the basis for constructing the anomaly judgment model. Based on these impact relationships, the system terminal analyzes the specific impact relationships between top events and basic events and the respective operation state data, and constructs corresponding training data. Subsequently, the obtained training data is used to train the structure of the anomaly judgment model to construct the required anomaly judgment model.

[0036] Furthermore, the present application provides a method for constructing an anomaly judgment model according to the impact relationships of the respective operation state data, and the method further includes:

[0037] Construct the structure of the anomaly judgment model according to the accident relationship tree, including a basic event anomaly judgment sub-model and a top event judgment sub-model;

[0038] Construct basic event training data and top event training data respectively according to the impact relationships of the respective operation state data;

[0039] Optionally, the system terminal uses the constructed accident relationship tree to guide the structure design of the anomaly judgment model. The accident relationship tree clearly shows the hierarchical structure and causal relationship of the accident, providing a logical framework for the anomaly judgment of basic events and top events. For basic events, the system terminal determines the operation state data directly related to the basic event according to the accident relationship tree and the definition of the basic event, and sets reasonable thresholds and judgment rules based on the impact relationship between the operation state data and the basic event. For example, when a data exceeds or is lower than a specific value, the basic event anomaly judgment sub-model should judge that the basic event is abnormal. Then, according to the set thresholds and rules, the structure of the basic event anomaly judgment sub-model is designed. This includes parts such as conditional judgment and logical operation. For top events, the system terminal sets the weights and priorities of different basic events according to the influence degree of the basic events on the top event in the accident relationship tree. Then, the structure of the top event judgment sub-model is initialized according to the set weights and priorities. This top event judgment sub-model not only considers the data anomalies directly affecting the top event but also comprehensively considers the anomaly judgment results of the basic events.

[0040] In the process of constructing these two sub-models, the system terminal identifies the operation status data indicators that are closely related to the basic events and top events based on the influence relationships of the operation status data. Analyze the influence degrees of these key indicators on the basic events and top events to determine their roles in the occurrence process of the events. Subsequently, according to historical experience, a clear label is determined for each basic event to indicate whether it has occurred. Then, associate the label of each basic event with the corresponding operation status data to form a data set containing event labels and related data. After that, also according to historical experience, a clear label is determined for each top event. Since the top event is often triggered by multiple basic events together, the system terminal associates the relevant basic event labels and operation status data with the top event. Then, integrate the label of the top event, the relevant basic event labels, and the operation status data to form a comprehensive training data set, that is, the top event training data.

[0041] In summary, an abnormal judgment model structure including basic event and top event judgment sub-models is constructed according to the accident relationship tree, and the corresponding training data is constructed by using the influence relationships of the operation status data, laying a foundation for the training and practical application of the model.

[0042] Learn by using the basic event training data and the top event training data respectively to obtain the basic event abnormal judgment sub-model and the top event judgment sub-model, and connect the basic event abnormal judgment sub-model and the top event judgment sub-model according to the hierarchical relationship of the accident relationship tree to obtain the abnormal judgment model.

[0043] Optionally, the system terminal divides the data to obtain two sets of training data. One is the basic event training data, which includes the operation status data related to the basic event and its corresponding labels; the other is the top event training data, which covers the basic event labels and operation status data related to the top event. Subsequently, the system terminal uses the basic event training data to train the constructed basic event anomaly judgment sub-model structure. By iteratively optimizing the parameters and structure of the basic event anomaly judgment sub-model, the basic event anomaly judgment sub-model structure can accurately identify the abnormal state of the basic event. When the maximum number of iterations is reached, the basic event anomaly judgment sub-model that has completed the current training is output. After that, the system terminal uses the same method to train the top event judgment sub-model structure with the top event training data, enabling the top event judgment sub-model to learn the complex relationship between the basic event and the top event, and to predict the occurrence of the top event based on the abnormal state of the basic event and other relevant features. Similarly, when the maximum number of iterations is reached, the top event judgment sub-model that has completed the current training is output. After obtaining these two sub-models, the system terminal connects them according to the hierarchical relationship of the accident relationship tree. The accident relationship tree describes the causal relationship between the basic event and the top event, so the system terminal determines the connection method between the sub-models based on this relationship. The output of the basic event anomaly judgment sub-model will be used as one of the inputs of the top event judgment sub-model, so that the top event judgment sub-model can make a judgment by comprehensively considering the abnormal state of the basic event. Finally, by connecting these two sub-models, the system terminal obtains a complete anomaly judgment model. This model can monitor the operation status data in real time and predict the occurrence of the basic event and the top event based on the changes in the data. It not only considers the anomaly judgment of a single event, but also comprehensively considers the causal relationship between events, improving the accuracy and reliability of the prediction. In summary, by using the basic event training data and the top event training data for learning and connecting the sub-models according to the accident relationship tree, an anomaly judgment model has been successfully constructed, providing a powerful tool for accident prevention and control.

[0044] Input the operation status data collected by the monitoring device into the anomaly judgment model for anomaly probability analysis, and output the anomaly judgment result;

[0045] In one embodiment, the system terminal obtains the operation status data collected in real time from the monitoring device. These data reflect the current working state of the energy storage system and are the key basis for judging whether there is an abnormality. Subsequently, these operation status data are input into the constructed abnormality judgment model. This model has learned the abnormality patterns of the basic events and the top event through training and can perform rapid and accurate abnormality probability analysis based on the input data. Inside the model, the data will sequentially pass through the basic event abnormality judgment sub-model and the top event judgment sub-model according to the hierarchical relationship of the accident relation tree. Among them, the basic event abnormality judgment sub-model will calculate the influence index of the input operation status data according to the learned knowledge and rules. The top event judgment sub-model will perform the weight calculation of each operation status data and the analysis of the accident occurrence probability, and calculate the corresponding abnormality probability. After that, the abnormality judgment model will output the occurrence probability of the abnormal event calculated by the top event judgment sub-model as the abnormality judgment result. This result can help the system terminal quickly understand the current state of the energy storage system, so as to take corresponding measures for risk control and prevention. In summary, by inputting the operation status data collected by the monitoring device into the abnormality judgment model for abnormality probability analysis, the abnormality judgment result can be quickly obtained, providing a strong guarantee for the safe operation of the energy storage system.

[0046] Furthermore, the present application provides a method for obtaining the basic event abnormality judgment sub-model and the top event judgment sub-model, and the method further includes:

[0047] Configuring a comparison matrix based on the influence relationship of the operation status data, and the comparison matrix is used to represent the importance of each status data for the occurrence of an abnormal accident;

[0048] Optionally, the system terminal first performs an influence degree conversion on the mutual influence relationship between each operation status data, quantifies the mutual influence relationship between each operation status data into a specific value for facilitating the subsequent construction of the matrix. Subsequently, the conversion results are sorted, and a comparison matrix is constructed based on this. Each row and each column of this matrix represents an operation status data, and the elements in the matrix represent the relative importance between these data. These importance values are determined by the conversion results. Through the comparison matrix, the system terminal can intuitively understand which status data are more important in an abnormal accident and which data have less influence. Such a matrix not only helps to deeply understand the mechanism of accident occurrence but also provides a strong basis for subsequent abnormality detection, prediction, and risk control. In summary, configuring the comparison matrix can help the system terminal comprehensively evaluate the importance of each operation status data in the occurrence of an abnormal accident, providing strong support for improving the safety and reliability of the energy storage system.

[0049] According to the comparison matrix, calculate the influence index of each operation status data, where the influence index is the sum of the matrix element values in the row or column where the operation status data is located;

[0050] Based on the influence index and the comparison matrix, establish a judgment matrix;

[0051] Optionally, after constructing the comparison matrix, the system terminal calculates the influence index of each operation status data. This index is obtained by adding the matrix element values in the row or column where the status data is located. The higher the influence index, the greater the importance of the status data in the overall system and the greater its contribution to the occurrence of abnormal accidents. Subsequently, the system terminal further constructs a judgment matrix based on the influence index and the comparison matrix through the judgment matrix formula. Among them, the judgment matrix formula is pre-constructed to calculate the judgment matrix factors and configure the judgment matrix element values. And the judgment matrix is constructed based on the configured judgment matrix element values, which not only includes the comparison results of the relative importance between status data, but also incorporates the information of the influence index, enabling the system terminal to more accurately evaluate the role and status of each status data in the occurrence of abnormal accidents. Through this process, the system terminal can not only understand the relative importance between each operation status data, but also quantify their contribution degrees to the occurrence of abnormal accidents. This provides a strong basis and guidance for actual safety management and risk control, helping the system terminal to more accurately identify key status data and evaluate the importance of each status data in the occurrence of abnormal accidents.

[0052] Calculate the weights of each operation status data according to the judgment matrix to obtain the weights of each data factor;

[0053] According to the weights of each data factor and the influence relationship, analyze the accident occurrence probability of the operation status data and output the occurrence probability of abnormal events.

[0054] Optionally, the system terminal standardizes the judgment matrix and then assigns weights to the corresponding operation status data using each element in the standardized judgment matrix. The weight reflects the relative importance of each status data in the whole. The greater the weight, the greater the influence of the status data in the occurrence of an accident. Subsequently, the system terminal conducts an in-depth accident occurrence probability analysis on the operation status data by combining the weights and influence relationships of each data factor. Specifically, for each operation status, the system terminal calculates the weighted sum of all its relevant data factors. The weighted sum is obtained by multiplying the value of each factor by its corresponding weight and then summing up all the products. This weighted sum reflects the potential risk of abnormal events occurring in this status. Subsequently, using the maximum value and the minimum value of the weighted sum obtained in advance, the weighted sum is standardized to obtain a standardized result between 0 and 1. Among them, the maximum value and the minimum value of the weighted sum are obtained by the system terminal through the above process by iteratively calculating a large number of historical sample data in advance. The system terminal outputs this standardized result as the occurrence probability of the abnormal event. This probability value reflects the likelihood of the abnormal event occurring under the current operation status data. In summary, by calculating the weights of each operation status data and combining their influence relationships for probability analysis, the occurrence risk of abnormal events can be evaluated more accurately, providing strong decision-making support for safety management.

[0055] Furthermore, the present application provides a method for configuring a comparison matrix based on the influence relationships of the above-mentioned operation status data, and the method further includes:

[0056] According to the influence relationships of the above-mentioned operation status data, perform influence degree conversion to obtain the influence degrees of each operation status data;

[0057] Optionally, the system terminal traces back the influence relationships of the obtained operation status data to understand in more detail the mutual influence relationships between each operation status data and determine the mutual influence intensity between each operation status data. Subsequently, based on historical experience and the mutual influence intensity between each operation status data, the system terminal assigns an influence score, that is, an influence degree, to each operation status data. The purpose of this step is to convert the original influence relationship between operation status data into a quantitative index that can reflect its influence degree, that is, the influence degree, so as to more accurately reflect the potential contribution of each data factor to the occurrence of abnormal events. These influence degree values can be used as an important basis for constructing the subsequent comparison matrix to help the system terminal better understand the operation status of the system and predict the occurrence probability of abnormal events.

[0058] According to the influence degrees of the above-mentioned operation status data, perform relationship sorting and configure matrix element values, where the first element value is that the jth element is more important than the kth element, the second element value is that the jth element is as important as the kth element, and the third element value is that the jth element is less important than the kth element;

[0059] Construct the comparison matrix based on the matrix element values.

[0060] Optionally, the system terminal compares the influence degrees between different element pairs to determine which pair of elements has a more significant mutual influence. According to the comparison results, the element pairs are sorted, that is, it is determined which element pairs have a greater influence on other elements. Based on the sorting result of the influence degree, the system terminal starts to configure the element values of the matrix. If the influence ability of the j-th element on the k-th element is significantly greater than the influence ability of the k-th element on the j-th element, that is, the j-th element is relatively more important, the system terminal sets the corresponding matrix element to the first element value. If the influence abilities between the j-th element and the k-th element are equivalent, that is, both have similar importance, then the system terminal sets the corresponding matrix element to the second element value. If the influence ability of the j-th element on the k-th element is significantly less than the influence ability of the k-th element on the j-th element, that is, the j-th element is less important than the k-th element, then the system terminal sets the corresponding matrix element to the third element value. Subsequently, based on these matrix element values, the system terminal constructs a complete comparison matrix by judging the matrix formula. This matrix reflects the relative importance between data factors. In summary, this process quantifies the influence degrees of each operating state data and converts them into matrix form to more intuitively represent and compare the relative importance between data factors. This is of great significance for subsequent system optimization, decision support, etc.

[0061] Furthermore, the present application provides a method for establishing a judgment matrix based on the influence index and the comparison matrix, and the method further includes:

[0062] According to the formula: , configure the element values of the judgment matrix and construct the judgment matrix;

[0063] wherein, is the judgment matrix factor, is the minimum value of the sum of the j-th row in the comparison matrix, is the maximum value of the sum of the j-th row in the comparison matrix, is the influence index of the j-th row, is the influence index of the k-th column.

[0064] Optionally, the process of constructing the judgment matrix calculates the element values of the judgment matrix through the judgment matrix formula. This formula is calculated based on the row sum, column sum in the comparison matrix, and the influence index of each row and column. The specific formula is as follows:

[0065] ;

[0066] wherein, It is the judgment matrix factor, that is, the element value in the j-th row and k-th column of the judgment matrix, representing the importance degree of the j-th operating state data relative to the k-th operating state data. Compare the minimum value of the sum of the elements in the j-th row of the matrix, which is used for calculation When it is used as a part of the denominator, it reflects the minimum value in the sum of the influence indexes of the elements in the j-th row. It is the maximum value of the sum of the elements in the j-th column of the comparison matrix, which is used for calculation When it is used as a part of the denominator, it reflects the maximum value in the sum of the influence indexes of the elements in the j-th column. It is the total influence index of the j-th row, that is, the sum of all elements in the j-th row of the comparison matrix, representing the comprehensive influence of the j-th operating state data in all comparisons. It is the total influence index of the k-th column, that is, the sum of all elements in the k-th column of the comparison matrix, representing the comprehensive influence of the k-th operating state data in all comparisons. By applying this formula, the system terminal can calculate multiple judgment matrix factors That is, the judgment matrix element value, so as to construct a more intuitive and easy-to-analyze judgment matrix, so as to better understand and optimize the relationship between the system operating state data.

[0067] Furthermore, the present application provides a method for calculating the weights of each operating state data according to the judgment matrix to obtain the weights of each data factor. The method further includes:

[0068] Standardize the judgment matrix so that the sum of the elements in each row of the judgment matrix is 1;

[0069] Use the elements of the matrix after the normalization process to calculate the weights of each operating state data to obtain the weights of each data factor.

[0070] Optionally, the system terminal performs a normalization process on the judgment matrix. The purpose of the normalization process is to make the sum of the elements in each row of the judgment matrix equal to 1, which can ensure that each operating state data is fairly considered when calculating the weights. The system terminal first calculates the sum of the elements in each row of the judgment matrix. Subsequently, divide the element value of each row by the sum of the elements in that row to achieve normalization. After completing the normalization process, the system terminal matches these normalized matrix elements with the corresponding operating state data. Then, use the value of the normalized element as the weight of the corresponding operating state data, so as to obtain the weights of each data factor. This is because the normalization process ensures that the sum of the element values in the same row is 1, so they have the property of weight distribution. By reasonably using these weight values, the operating state of the energy storage system can be grasped more accurately, and the stability and efficiency of the energy storage system can be improved.

[0071] Furthermore, the present application provides inputting the operating status data collected by the monitoring device into the abnormality judgment model to perform abnormality probability analysis and output an abnormality judgment result, and the method also includes:

[0072] The power, charge and discharge status, temperature, voltage, and insulation resistance impedance values ​​acquired through monitoring are input into the abnormality judgment model, and the basic event abnormality probability analysis is performed through the basic event abnormality judgment sub-model to obtain the basic abnormal events and their occurrence probabilities;

[0073] Input the basic abnormal events and their occurrence probabilities into the top event judgment sub-model, perform top event occurrence probability analysis, and determine top abnormal events and their occurrence probabilities;

[0074] The basic abnormal events and their occurrence probabilities, the top abnormal events and their occurrence probabilities are output as abnormality judgment results.

[0075] Optionally, the system terminal inputs the key data such as the monitored power, charge and discharge state, temperature, voltage and insulation resistance impedance value into the abnormal judgment model. This model contains the basic event abnormal judgment sub-model and the top event judgment sub-model, which will analyze these data in turn. Specifically, within the basic event abnormal judgment sub-model, the key data such as the input power, charge and discharge state, temperature, voltage and insulation resistance impedance value are subjected to matrix creation and other operations by the aforementioned method of generating the probability of occurrence of abnormal events, and the basic abnormal events and the probability of occurrence of basic abnormal events are determined. Subsequently, the basic event abnormal judgment sub-model transmits the generated basic abnormal events and the probability of occurrence of basic abnormal events to the top event judgment sub-model for further analysis; after receiving these data, the top event judgment sub-model also performs the same operation as above to determine the top abnormal events and the probability of occurrence of top abnormal events. After that, the abnormal judgment model outputs the basic abnormal events and their probability of occurrence, the top abnormal events and their probability of occurrence as the result of the entire abnormal judgment. In this way, the system terminal obtains a comprehensive and detailed abnormal analysis report, which can better understand the operating status of the energy storage system and discover potential problems in time.

[0076] Generate warning information according to the abnormality judgment result, load the abnormality judgment result into the historical record library, and update the abnormality record database.

[0077] In one embodiment, the system terminal further generates warning information based on the basic abnormal events and their occurrence probabilities, as well as the top abnormal events and their occurrence probabilities in the abnormal judgment results. These warning information will target the possible abnormal events, reminding the user to pay attention and take corresponding measures. By doing so, potential energy storage system problems can be detected and addressed in a timely manner, ensuring the stable operation of the energy storage system. At the same time, the system terminal also loads these abnormal judgment results into the historical record library to update the historical data. In this way, the past abnormal records can be reviewed at any time to understand the past operating status of the energy storage system. By continuously updating and improving the abnormal record database, a complete abnormal historical database can gradually be formed, making the extracted abnormal accident case set more and more perfect, providing strong support for the long-term stable operation of the energy storage system.

[0078] Further, the present application provides a method for generating warning information according to the abnormal judgment result, loading the abnormal judgment result into the historical record library, and updating the abnormal record database. The method further includes:

[0079] Obtain a warning configuration database, which includes abnormal event types, warning probability thresholds, and warning methods;

[0080] Match the basic abnormal events and top abnormal events with the abnormal event types, and when the warning probability threshold is met, generate warning information according to the warning method;

[0081] Generate warning record data according to the generation time of the warning information and store it in the abnormal record database, which is used to record and query the warning information.

[0082] Optionally, the system terminal establishes a connection with the warning configuration database through the interface of the embedded platform. This warning configuration database includes various possible abnormal event types, corresponding warning probability thresholds, and warning methods to be taken when an abnormality occurs. Subsequently, the system terminal matches the analyzed basic abnormal events and top abnormal events with the abnormal event types in the warning configuration database. If the occurrence probability of an abnormal event exceeds the preset warning probability threshold, then the system terminal will generate warning information according to the warning method specified in the warning configuration database. The warning information includes alarm sounds, SMS notifications, email reminders, etc., and the specific method depends on the settings in the warning configuration. While generating the warning information, the system terminal also generates corresponding warning record data according to the generation time of the warning information and stores these data in the abnormal record database. Through this process, the system terminal can not only detect and handle abnormal events in the system in a timely manner, but also completely record the detailed information of each warning, providing strong support for the long-term stable operation of the energy storage system.

[0083] In summary, the embodiments of the present application have at least the following technical effects:

[0084] In the embodiments of the present application, a monitoring device is connected through an embedded platform to obtain key operation status data such as power, charge and discharge status, temperature, voltage, and insulation resistance in real time. Based on these data, an abnormal accident case set is extracted from the historical record library, and the influence relationship between the operation status data is deeply analyzed. Subsequently, an abnormal judgment model is constructed using these influence relationships. The model includes a basic event abnormal judgment sub-model and a top event judgment sub-model, which can accurately evaluate the abnormal probability of each operation status data. During the early warning process, the real-time monitoring data is input into the abnormal judgment model for abnormal probability analysis, so as to output the basic abnormal event and the top abnormal event and their occurrence probabilities. Combining the abnormal event type, early warning probability threshold, and early warning method in the early warning configuration database, corresponding early warning information can be automatically generated. At the same time, the abnormal judgment result will be loaded into the historical record library to update the abnormal record database for subsequent recording, querying, and analysis of the early warning information. Through this method, the abnormal status of the energy storage system can be detected and processed in a timely manner, improving the safety and reliability of the energy storage system. These technical effects together solve the technical problem in the existing early warning that due to insufficient data analysis depth methods and low accuracy of the early warning model, the abnormal status of the energy storage system cannot be discovered in a timely and accurate manner. It realizes the technical effect that through the combination of real-time monitoring data and the abnormal judgment model, the abnormal status of the energy storage system can be timely warned, effectively reducing the false negative rate of the early warning and improving the accuracy and timeliness of the early warning.

[0085] It should be noted that the above sequence of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of this specification have been described. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired result. In some embodiments, multi-tasking and parallel processing are also possible or may be advantageous.

[0086] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0087] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A method for early warning of abnormal state of an energy storage system, characterized in that: The method is applied to an energy storage system, wherein the energy storage system comprises an embedded platform, including: Connect monitoring equipment through the embedded platform interface to obtain operating status data, including power, charge and discharge status, temperature, voltage, and insulation resistance impedance value; Connecting to the history record library, extracting an abnormal accident case set based on the electric quantity, charge and discharge state, temperature, voltage, and insulation resistance impedance value, wherein the abnormal accident case set includes one or more abnormal operation state data of the electric quantity, charge and discharge state, temperature, voltage, and insulation resistance impedance value; Analyzing each operating status data of the abnormal accident case set to determine the influence relationship of each operating status data; Constructing an abnormality judgment model according to the influence relationship of each operating status data; Input the operating status data collected by the monitoring device into the abnormality judgment model to perform abnormality probability analysis and output the abnormality judgment result; Generate warning information according to the abnormal judgment result, load the abnormal judgment result into the historical record library, and update the abnormal record database; The analysis of each operating status data of the abnormal accident case set to determine the influence relationship of each operating status data includes: Clustering the abnormal accident case set into accident types to obtain a multi-accident set; Performing longitudinal comparison on the operation status data of the multiple accident sets respectively to obtain a change relationship of the operation status data, and fitting an influence relationship between the operation status data and the accident type based on the change relationship of the operation status data; Perform accident concurrent status analysis on the multiple accident sets, determine the types of concurrent accidents, construct an accident relationship tree, and determine top events and basic events; According to the influence relationship, the influence relationship of each running status data is fitted for the top event and the basic event respectively, and the influence relationship between the abnormal event and each running status data is established.

2. The method according to claim 1, characterized in that According to the influence relationship of each operating status data, an abnormality judgment model is constructed, including: According to the accident relationship tree, an abnormality judgment model structure is constructed, including a basic event abnormality judgment sub-model and a top event judgment sub-model; According to the influence relationship of each running status data, basic event training data and top event training data are respectively constructed; The basic event training data and the top event training data are used for learning respectively to obtain the basic event abnormality judgment submodel and the top event judgment submodel, and the basic event abnormality judgment submodel and the top event judgment submodel are connected according to the hierarchical relationship of the accident relationship tree to obtain the abnormality judgment model.

3. The method according to claim 2, characterized in that Obtaining the basic event abnormality judgment sub-model and the top event judgment sub-model includes: A comparison matrix is ​​configured based on the influence relationship of each operating state data, wherein the comparison matrix is ​​used to characterize the importance of each state data to the occurrence of an abnormal accident; Calculate the influence index of each running status data according to the comparison matrix, where the influence index is the sum of the matrix element values ​​of the row or column where the running status data is located; Based on the influence index and the comparison matrix, a judgment matrix is ​​established; Calculate the weight of each running status data according to the judgment matrix to obtain the weight of each data factor; According to the weights and influence relationships of the data factors, the accident probability analysis is performed on the operating status data, and the probability of occurrence of abnormal events is output.

4. The method according to claim 3, characterized in that The comparison matrix is ​​configured based on the influence relationship of each operation status data, including: According to the influence relationship of each running status data, influence degree conversion is performed to obtain the influence degree of each running status data; Sort the relationships according to the influence of each running status data, and configure matrix element values, wherein the first element value is that the j-th element is more important than the k-th element, the second element value is that the j-th element is as important as the k-th element, and the third element value is that the j-th element is less important than the k-th element; The comparison matrix is ​​constructed based on the matrix element values.

5. The method according to claim 4, characterized in that Based on the influence index and comparison matrix, a judgment matrix is ​​established, including: According to the formula: , configure the judgment matrix element values ​​and construct the judgment matrix; in, is the judgment matrix factor, To compare the minimum value of the sum of the jth row in the matrix, To compare the maximum value of the sum of the jth row in the matrix, is the influence index of the jth row, is the influence index of the kth column.

6. The method according to claim 3, characterized in that The weight of each operating status data is calculated according to the judgment matrix to obtain the weight of each data factor, including: Standardizing the judgment matrix so that the sum of the elements of each row of the judgment matrix is ​​1; The standardized matrix elements are used to perform weight calculation on the operating status data to obtain the weight of each data factor.

7. The method according to claim 3, characterized in that The operation status data collected by the monitoring device is input into the abnormality judgment model to perform abnormality probability analysis, and the abnormality judgment result is output, including: The power, charge and discharge status, temperature, voltage, and insulation resistance impedance values ​​acquired through monitoring are input into the abnormality judgment model, and the basic event abnormality probability analysis is performed through the basic event abnormality judgment sub-model to obtain the basic abnormal events and their occurrence probabilities; Input the basic abnormal events and their occurrence probabilities into the top event judgment sub-model, perform top event occurrence probability analysis, and determine top abnormal events and their occurrence probabilities; The basic abnormal events and their occurrence probabilities, the top abnormal events and their occurrence probabilities are output as abnormality judgment results.

8. The method according to claim 7, characterized in that Generating warning information according to the abnormality judgment result, and loading the abnormality judgment result into the historical record database, and updating the abnormality record database, including: Obtain the warning configuration database, which includes abnormal event types, warning probability thresholds, and warning methods; The basic abnormal event, the top abnormal event and the abnormal event type are matched, and when the warning probability threshold is met, warning information is generated according to the warning method; According to the generation time of the warning information, the warning record data is generated and stored in the abnormal record database, and the abnormal record database is used to record and query the warning information.

Images