Fault detection method and detection device for energy storage power station

Abstract

The invention discloses a fault detection method and device for an energy storage power station, and belongs to the technical field of energy storage, and the method comprises the steps: obtaining the charging and discharging control parameters of a first energy storage unit of the energy storage power station; receiving environment temperature monitoring information and environment humidity monitoring information; battery voltage fluctuation prediction is carried out according to the parameters and the information, and a voltage fluctuation prediction curve is obtained; performing battery temperature fluctuation prediction to obtain a temperature fluctuation prediction curve; comparing the prediction curve with the monitoring curve to obtain a voltage fluctuation deviation coefficient and a temperature fluctuation deviation coefficient; and when the deviation coefficient exceeds a threshold value, generating a fault detection instruction to execute dynamic fault detection. By analyzing the operation data of the energy storage power station, early-stage high-precision detection of potential faults of the battery is realized, and targeted operation and maintenance can be carried out on the battery with hidden dangers before problem deterioration, so that the operation and maintenance cost of the energy storage power station is greatly reduced, and the safety and reliability of the energy storage power station are improved.

Description

Technical Field

[0001] The present invention relates to the field of energy storage, and in particular to a fault detection method and a fault detection device for an energy storage power station. Background Art

[0002] As a key component of smart grids, energy storage power stations play an important role in smoothing fluctuations in renewable energy power generation, peak-shaving and valley-filling, and improving grid reliability. However, energy storage batteries will inevitably experience various faults during long-term operation, affecting the normal operation of power stations. Therefore, timely and accurate detection and diagnosis of energy storage battery faults are crucial to ensure the safe and stable operation of energy storage power stations. Existing energy storage power station fault detection mainly relies on threshold judgments of battery voltage, current, temperature and other parameters, and can only identify obvious anomalies that have already occurred, such as battery overcharge, overdischarge, overheating, etc., but cannot detect degradation or hidden dangers occurring inside the battery. If these potential problems are not discovered and handled in a timely manner, they will gradually deteriorate, eventually leading to battery failure and even safety accidents. At this point, the cost of repairing or replacing the battery will increase significantly. Summary of the invention

[0003] The present application provides a fault detection method and a detection device for an energy storage power station, aiming to solve the technical problem that the prior art can only identify obvious abnormal faults in the energy storage power station, but cannot perform targeted operation and maintenance on batteries with potential hidden dangers, resulting in high maintenance costs after the problem worsens.

[0004] In view of the above problems, the present application provides a fault detection method and a detection device for an energy storage power station.

[0005] In a first aspect disclosed in the present application, a fault detection method for an energy storage power station is provided, the method comprising: obtaining a charge and discharge control parameter of a first energy storage unit of the energy storage power station; interacting with an environmental sensor to receive environmental temperature monitoring information and environmental humidity monitoring information; predicting battery voltage fluctuations according to the charge and discharge control parameters, the environmental temperature monitoring information and the environmental humidity monitoring information to obtain a voltage fluctuation prediction curve; predicting battery temperature fluctuations according to the charge and discharge control parameters, the environmental temperature monitoring information and the environmental humidity monitoring information to obtain a temperature fluctuation prediction curve; obtaining a voltage fluctuation monitoring curve and a temperature fluctuation monitoring curve; comparing the voltage fluctuation prediction curve with the voltage fluctuation monitoring curve to obtain a voltage fluctuation deviation coefficient, and comparing the temperature fluctuation prediction curve with the temperature fluctuation monitoring curve to obtain a temperature fluctuation deviation coefficient, wherein the voltage fluctuation deviation coefficient represents a proportion of time periods greater than or equal to a voltage deviation threshold, and the temperature fluctuation deviation coefficient represents a proportion of time periods greater than or equal to a temperature deviation threshold; when the voltage fluctuation deviation coefficient is greater than or equal to a voltage fluctuation deviation coefficient threshold, or / and the temperature fluctuation deviation coefficient is greater than or equal to a temperature fluctuation deviation coefficient threshold, generating a first energy storage unit fault detection instruction to perform dynamic fault detection.

[0006] Another aspect disclosed in the present application provides a fault detection device for an energy storage power station, the device comprising: a parameter acquisition module, used to obtain the charge and discharge control parameters of the first energy storage unit of the energy storage power station; an environmental monitoring module, used to interact with the environmental sensor and receive the environmental temperature monitoring information and the environmental humidity monitoring information; a voltage prediction module, used to predict the battery voltage fluctuation according to the charge and discharge control parameters, the environmental temperature monitoring information and the environmental humidity monitoring information, and obtain a voltage fluctuation prediction curve; a temperature prediction module, used to predict the battery temperature fluctuation according to the charge and discharge control parameters, the environmental temperature monitoring information and the environmental humidity monitoring information, and obtain a temperature fluctuation prediction curve; a measured data acquisition module, used to obtain A voltage fluctuation monitoring curve and a temperature fluctuation monitoring curve; a deviation analysis module, used to compare the voltage fluctuation prediction curve and the voltage fluctuation monitoring curve to obtain a voltage fluctuation deviation coefficient, and to compare the temperature fluctuation prediction curve and the temperature fluctuation monitoring curve to obtain a temperature fluctuation deviation coefficient, wherein the voltage fluctuation deviation coefficient represents the proportion of moments greater than or equal to a voltage deviation threshold, and the temperature fluctuation deviation coefficient represents the proportion of moments greater than or equal to a temperature deviation threshold; a fault detection module, used to generate a first energy storage unit fault detection instruction to perform dynamic fault detection when the voltage fluctuation deviation coefficient is greater than or equal to a voltage fluctuation deviation coefficient threshold, or / and the temperature fluctuation deviation coefficient is greater than or equal to a temperature fluctuation deviation coefficient threshold.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The charging and discharging control parameters of the first energy storage unit of the energy storage power station are obtained as the key parameters of the operating status of the energy storage unit, providing basic data for subsequent fault detection; interacting with environmental sensors, receiving ambient temperature monitoring information and ambient humidity monitoring information, considering the impact of environmental factors on battery performance, making fault detection more comprehensive and accurate; predicting battery voltage fluctuations based on the charging and discharging control parameters, ambient temperature monitoring information and ambient humidity monitoring information, obtaining a voltage fluctuation prediction curve, and predicting the voltage change trend of the battery by comprehensively analyzing the operating parameters and environmental factors, providing a reference for fault diagnosis; predicting battery temperature fluctuations based on the charging and discharging control parameters, ambient temperature monitoring information and ambient humidity monitoring information, obtaining a temperature fluctuation prediction curve, and predicting the temperature change trend of the battery by comprehensively analyzing the operating parameters and environmental factors, providing a reference for fault diagnosis; obtaining a voltage fluctuation monitoring curve and a temperature fluctuation monitoring curve for comparison with the prediction curve; comparing the voltage fluctuation prediction curve and the voltage fluctuation monitoring curve to obtain the voltage fluctuation deviation coefficient, and comparing the temperature fluctuation prediction curve with the voltage fluctuation monitoring curve to obtain the voltage fluctuation deviation coefficient. The voltage fluctuation monitoring curve and the temperature fluctuation monitoring curve are measured to obtain the temperature fluctuation deviation coefficient. The deviation coefficient between the predicted curve and the measured curve is calculated to quantify the difference between the actual operating state of the battery and the expected state, so as to provide a basis for fault judgment. When the voltage fluctuation deviation coefficient is greater than or equal to the voltage fluctuation deviation coefficient threshold, or / and the temperature fluctuation deviation coefficient is greater than or equal to the temperature fluctuation deviation coefficient threshold, a first energy storage unit fault detection instruction is generated to execute dynamic fault detection. According to the comparison result between the deviation coefficient and the preset threshold, it is determined whether the battery has a fault, and a further fault detection process is started, so as to realize a technical solution for early detection of potential faults, solve the technical problem that the prior art can only identify obvious abnormal faults of energy storage power stations, but cannot carry out targeted operation and maintenance of batteries with potential hidden dangers, resulting in high maintenance costs after the problem worsens. By analyzing the operation data of the energy storage power station, early high-precision detection of potential battery faults can be realized, and targeted operation and maintenance can be carried out on batteries with hidden dangers before the problem worsens, thereby greatly reducing the operation and maintenance costs of the energy storage power station and improving its safety and reliability.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic flow chart of a fault detection method for an energy storage power station is provided for an embodiment of the present application;

[0011] Figure 2A structural schematic diagram of a fault detection device for an energy storage power station is provided for an embodiment of the present application.

[0012] Explanation of the reference numerals: parameter acquisition module 11 , environment monitoring module 12 , voltage prediction module 13 , temperature prediction module 14 , measured data acquisition module 15 , deviation analysis module 16 , fault detection module 17 . DETAILED DESCRIPTION

[0013] The overall idea of ​​the technical solution provided by this application is as follows:

[0014] The embodiment of the present application provides a fault detection method and detection device for an energy storage power station. First, the charge and discharge control parameters of the first energy storage unit in the energy storage power station are obtained, and the ambient temperature and humidity monitoring information is received by interacting with the environmental sensor to fully grasp the operating state and working environment of the battery. Then, according to the obtained charge and discharge control parameters and environmental monitoring information, the battery voltage fluctuation and temperature fluctuation are predicted respectively, and the corresponding prediction curves are obtained as a reference for judging whether the battery is abnormal. At the same time, the voltage and temperature fluctuation monitoring curves during the actual operation of the battery are obtained. Next, by comparing the prediction curve and the monitoring curve, the voltage and temperature deviation coefficients are calculated to evaluate the degree of difference between the actual operating state of the battery and the expected state. Afterwards, when the deviation coefficient exceeds the preset threshold, it is determined that the energy storage unit has a potential fault, a fault detection instruction is generated, and a further dynamic fault detection process is started to achieve early detection and diagnosis of potential faults. Compared with the prior art, it can significantly improve the fault detection efficiency and accuracy of the energy storage power station, minimize the safety risks and economic losses caused by battery failures, and ensure the safe and stable operation of the energy storage power station.

[0015] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0016] Embodiment 1, as Figure 1 As shown, an embodiment of the present application provides a fault detection method for an energy storage power station, the method comprising:

[0017] S100: Obtaining charging and discharging control parameters of a first energy storage unit of an energy storage power station.

[0018] Specifically, during the operation of the energy storage power station, the energy storage unit will be affected by various control parameters during the charging and discharging process. These control parameters include but are not limited to charging current, charging voltage, charging time, discharging current, discharging voltage, discharging time, etc. By obtaining the charging and discharging control parameters of the first energy storage unit, important data is provided for subsequent fault detection and analysis. Among them, the first energy storage unit is any energy storage unit in the energy storage power station.

[0019] Among them, the charge and discharge control parameters can be obtained in various ways, such as directly reading from the control system of the energy storage power station, measuring in real time through sensors, etc. The obtained charge and discharge control parameters are stored and represented in various forms, such as numerical values, curves, vectors, etc., to facilitate subsequent processing and analysis.

[0020] By obtaining the charge and discharge control parameters of the first energy storage unit, a data foundation is laid for subsequent fault detection and analysis.

[0021] S200: interact with environmental sensors to receive environmental temperature monitoring information and environmental humidity monitoring information.

[0022] Specifically, the operation of energy storage power stations is not only affected by the charging and discharging control parameters, but also closely related to external environmental conditions. Among them, temperature and humidity are two important environmental parameters that have a significant impact on the performance and life of energy storage units.

[0023] By interacting with the environmental sensor corresponding to the first energy storage unit, the environmental temperature monitoring information and environmental humidity monitoring information of the environment in which the first energy storage unit is located are obtained in real time. Among them, the environmental sensors are selected and deployed according to actual needs, for example, using temperature sensors and humidity sensors to measure the environmental temperature and humidity respectively, or using integrated temperature and humidity sensors to measure the two parameters simultaneously. Through the interactive environmental sensor, the environmental temperature monitoring information and the environmental humidity monitoring information are received to provide basic information for evaluating the health status and failure risk of the first energy storage unit.

[0024] By interacting with the environmental sensor, the ambient temperature monitoring information and ambient humidity monitoring information of the first energy storage unit are obtained. Together with the charge and discharge control parameters, this information provides comprehensive data support for subsequent fault detection analysis, which helps to improve the accuracy and reliability of fault detection.

[0025] S300: Predicting battery voltage fluctuation according to the charge and discharge control parameters, the ambient temperature monitoring information, and the ambient humidity monitoring information to obtain a voltage fluctuation prediction curve.

[0026] Specifically, in order to predict the battery voltage fluctuation, first, a battery voltage fluctuation prediction model is established. The model uses machine learning algorithms such as support vector machines, neural networks, long short-term memory networks, etc. By training a large amount of historical operation data, a mapping relationship between input parameters and voltage fluctuations is established. In the process of model training, historical operation data are collected and preprocessed, including charge and discharge control parameters, ambient temperature monitoring information, ambient humidity monitoring information, and actual voltage measurements at the corresponding time. The collected data is preprocessed by denoising, normalization, and other operations to improve data quality; the processed data is divided into a training set and a validation set, where the training set is used for model training and parameter adjustment, and the validation set is used to evaluate the performance of the model; according to the selected machine learning algorithm, the model structure and hyperparameters are designed; by inputting the input data and target output of the training set into the model, the model parameters are iteratively adjusted using the optimization algorithm to minimize the error between the predicted output and the actual voltage value; after each training cycle, the performance of the model is evaluated using the validation set, and the model is further optimized based on the evaluation results; after multiple iterations of training and optimization, a battery voltage fluctuation prediction model is obtained. Afterwards, the real-time acquired charge and discharge control parameters, ambient temperature monitoring information, and ambient humidity monitoring information are input into the battery voltage fluctuation prediction model to obtain a voltage fluctuation prediction curve.

[0027] By establishing a battery voltage fluctuation prediction model and using the real-time acquired charging and discharging control parameters, ambient temperature monitoring information, and ambient humidity monitoring information for prediction, a voltage fluctuation prediction curve is obtained, which provides important data support for subsequent fault detection, helps to timely discover and diagnose potential battery problems, and improve the reliability and safety of energy storage power stations.

[0028] S400: Predicting battery temperature fluctuations according to the charge and discharge control parameters, the ambient temperature monitoring information, and the ambient humidity monitoring information to obtain a temperature fluctuation prediction curve.

[0029] Specifically, in order to predict battery temperature fluctuations according to charge and discharge control parameters, ambient temperature monitoring information, and ambient humidity monitoring information, a battery temperature fluctuation prediction model based on a long short-term memory network is first established. During the training phase of the model, sufficient historical operation data is collected, including charge and discharge control parameters, ambient temperature monitoring information, ambient humidity monitoring information, and actual battery temperature measurements at the corresponding time. The collected data is preprocessed, such as time alignment, outlier removal, data normalization, etc., to improve data quality and consistency. The preprocessed historical data is arranged in chronological order to construct a time series data set. Then, the time series data set is divided into a training set, a validation set, and a test set. The training set is used for model training and parameter optimization, the validation set is used for model hyperparameter selection and performance evaluation, and the test set is used to evaluate the final performance of the model. Subsequently, the training set data is used to adjust the weights and biases of the model through the back propagation algorithm and optimizer to minimize the mean square error between the predicted temperature and the actual temperature. After each training cycle, the validation set is used to evaluate the performance of the model, such as calculating the mean absolute error, root mean square error, and other indicators. According to the verification results, the hyperparameters of the model, such as the number of hidden units and the learning rate, are adjusted to improve the performance and generalization ability of the model. After multiple iterations of training and tuning, a battery temperature fluctuation prediction model is obtained. By applying the model to the real-time acquired charge and discharge control parameters, ambient temperature monitoring information, and ambient humidity monitoring information, the predicted value of battery temperature fluctuation in the future period can be obtained. The predicted values ​​of battery temperature fluctuation in the future period are connected into a curve to generate a temperature fluctuation prediction curve.

[0030] By predicting the battery temperature fluctuation, a temperature fluctuation prediction curve is obtained. This prediction result, together with the voltage fluctuation prediction curve, provides important data support for fault detection in energy storage power stations, which helps to improve the timeliness and accuracy of fault diagnosis.

[0031] S500: Obtain a voltage fluctuation monitoring curve and a temperature fluctuation monitoring curve.

[0032] Specifically, during the operation of the first energy storage unit, the voltage and temperature data of the battery are collected in real time through the voltage sensor and temperature sensor configured on the first energy storage unit. For voltage fluctuation monitoring, the voltage sensor continuously collects the actual output voltage value of the battery at a certain sampling frequency (such as 10 times per second), arranges the collected voltage values ​​in chronological order, and forms a continuous voltage fluctuation monitoring curve, which reflects the changes in the battery voltage during the actual operation process, including the voltage fluctuation amplitude, frequency, trend and other characteristics. Similarly, for temperature fluctuation monitoring, the temperature sensor continuously collects the battery temperature value at a certain sampling frequency, arranges the collected temperature values ​​in chronological order, and forms a continuous temperature fluctuation monitoring curve, which reflects the changes in the battery temperature during the actual operation process, including the temperature fluctuation amplitude, rate, trend and other characteristics.

[0033] After obtaining the voltage fluctuation monitoring curve and the temperature fluctuation monitoring curve, the curve data is preprocessed to improve data quality and availability. For example, identify and remove abnormal value points in the curve, such as sudden changes in voltage or temperature, exceeding the normal range, etc., to avoid interference with subsequent analysis; smooth the curve to remove high-frequency noise and jitter, improve the smoothness and continuity of the curve, and facilitate the extraction of key features; align the voltage fluctuation monitoring curve and the temperature fluctuation monitoring curve on the time axis to ensure that the data points of the two curves correspond one to one in time, which is convenient for joint analysis. After preprocessing, the obtained voltage fluctuation monitoring curve and temperature fluctuation monitoring curve will be used as the basic data for fault detection analysis. By comparing and analyzing the actual monitoring curve with the predicted curve, abnormal deviations in the battery operating status can be found, and potential fault signs can be detected in time.

[0034] By acquiring the actual operating data of the first energy storage unit, a voltage fluctuation monitoring curve and a temperature fluctuation monitoring curve are generated, which truly reflect the voltage and temperature characteristics of the battery and provide a reliable data basis for subsequent fault detection and analysis.

[0035] S600: Compare the voltage fluctuation prediction curve with the voltage fluctuation monitoring curve to obtain a voltage fluctuation deviation coefficient, and compare the temperature fluctuation prediction curve with the temperature fluctuation monitoring curve to obtain a temperature fluctuation deviation coefficient, wherein the voltage fluctuation deviation coefficient represents the proportion of moments greater than or equal to a voltage deviation threshold, and the temperature fluctuation deviation coefficient represents the proportion of moments greater than or equal to a temperature deviation threshold.

[0036] Specifically, first, the obtained voltage fluctuation prediction curve is compared with the obtained voltage fluctuation monitoring curve. Since the two curves are aligned on the time axis, the difference between the predicted value and the actual value can be compared point by point. For each moment, the absolute deviation between the predicted voltage value and the actual voltage value is calculated to obtain a voltage deviation value. Next, a voltage deviation threshold is set to determine whether the voltage deviation is abnormal. Among them, the voltage deviation threshold is determined according to the parameters such as the battery model, capacity, rated voltage, and the statistical analysis of historical operation data. Then, the number of moments when the voltage deviation value is greater than or equal to the voltage deviation threshold during the entire monitoring period is counted, and its proportion of the total number of monitoring moments is calculated to obtain the voltage fluctuation deviation coefficient, which reflects the severity and duration of the battery voltage deviation from the expected value. Similarly, for the calculation of the temperature fluctuation deviation coefficient, the obtained temperature fluctuation prediction curve is compared with the obtained temperature fluctuation monitoring curve point by point to calculate the temperature deviation value at each moment. The temperature deviation threshold is set, the number of moments when the temperature deviation value is greater than or equal to the threshold is counted, and its proportion of the total number of monitoring moments is calculated to obtain the temperature fluctuation deviation coefficient. Among them, the larger the deviation coefficient, the more serious the deviation between the actual operating state of the battery and the expected state, and the higher the possibility of failure.

[0037] By comparing the prediction curve and the monitoring curve, calculating the voltage fluctuation deviation coefficient and the temperature fluctuation deviation coefficient, the degree to which the actual operating state of the battery deviates from the expected state is quantified, which provides an important basis for subsequent fault diagnosis and maintenance decisions, helps to timely discover and deal with abnormal battery conditions, and improve the reliability and safety of energy storage power stations.

[0038] S700: When the voltage fluctuation deviation coefficient is greater than or equal to the voltage fluctuation deviation coefficient threshold, or / and the temperature fluctuation deviation coefficient is greater than or equal to the temperature fluctuation deviation coefficient threshold, a first energy storage unit fault detection instruction is generated to perform dynamic fault detection.

[0039] Specifically, first, determine whether the voltage fluctuation deviation coefficient is greater than or equal to the voltage fluctuation deviation coefficient threshold. If so, it indicates that the voltage fluctuation of the first energy storage unit is abnormal, and there may be voltage-related faults, such as overcharging, over-discharging, short circuit, etc. At this time, a fault detection instruction for voltage abnormality is generated as the first energy storage unit fault detection instruction, and the corresponding fault diagnosis program is triggered. Similarly, determine whether the temperature fluctuation deviation coefficient is greater than or equal to the temperature fluctuation deviation coefficient threshold. If so, it indicates that the temperature fluctuation of the first energy storage unit is abnormal, and there may be temperature-related faults, such as overheating, poor heat dissipation, thermal runaway, etc. At this time, a fault detection instruction for temperature abnormality is generated as the first energy storage unit fault detection instruction, and the corresponding fault diagnosis program is triggered. Among them, voltage abnormality and temperature abnormality may occur at the same time, or only one of them may occur.

[0040] After the fault detection instruction is generated, dynamic fault detection will be performed. First, the key parameters of the energy storage unit, such as voltage, current, temperature, etc., are continuously collected to record real-time operation data in a high-frequency and high-precision manner; then, key indicators that can reflect fault characteristics are extracted from the collected real-time data, such as voltage mutation, temperature rise, internal resistance increase, etc.; then, the extracted fault characteristics are input into the pre-trained fault diagnosis model, such as decision tree, support vector machine, neural network, etc., to distinguish and classify the fault type; then, according to the diagnosis results, combined with the topological structure and electrical connection of the energy storage unit, the specific location of the fault is determined, such as a battery module, a battery string, etc.; then, according to the amplitude and duration of the fault characteristics, the severity of the fault is evaluated, such as mild, moderate, severe, etc., to provide a basis for subsequent maintenance decisions; at the same time, according to the results of the fault diagnosis, fault alarm information is generated to notify the operation and maintenance personnel to handle it, and the fault event and diagnosis results are recorded in the log for subsequent analysis and tracing. The process of dynamic fault detection is real-time and continuous, which can quickly respond to changes in the state of the energy storage unit and promptly discover and diagnose faults. Through dynamic fault detection, the impact of faults on energy storage power stations can be minimized and the reliability of energy storage power stations can be improved.

[0041] By generating fault detection instructions based on the abnormal situation of the deviation coefficient and triggering the dynamic fault detection process, the fault of the energy storage unit can be discovered and located in time, effectively improving the operating safety and reliability of the energy storage power station, reducing the economic losses and safety risks caused by faults, and reducing the operation and maintenance costs of the energy storage power station.

[0042] Furthermore, the embodiment of the present application also includes:

[0043] S800: When the voltage fluctuation deviation coefficient is less than the voltage fluctuation deviation coefficient threshold, and the temperature fluctuation deviation coefficient is less than the temperature fluctuation deviation coefficient threshold, obtaining voltage variance fluctuation curves of a plurality of voltage fluctuation prediction curves of the first energy storage unit;

[0044] S900: Counting the proportion of the curve length greater than or equal to the balance variance threshold in the voltage variance fluctuation curve, and setting it as the voltage balance abnormality coefficient;

[0045] S1000: When the voltage balance abnormality coefficient is greater than or equal to the voltage balance abnormality coefficient threshold, a second energy storage unit fault detection instruction is generated to perform dynamic fault detection.

[0046] In a feasible implementation, when it is judged that the voltage fluctuation deviation coefficient is less than the voltage fluctuation deviation coefficient threshold, and the temperature fluctuation deviation coefficient is less than the temperature fluctuation deviation coefficient threshold, it means that the voltage fluctuation and temperature fluctuation of the first energy storage unit are within the normal range, and there is no obvious overcharge, over-discharge, short circuit, overheating and other faults. However, this does not mean that the first energy storage unit is completely healthy, and it is necessary to further check whether there is a problem of abnormal balance in the first energy storage unit. To this end, several voltage fluctuation prediction curves of the first energy storage unit are obtained. These curves are obtained by predicting the voltage fluctuation of the first energy storage unit in different time periods. Then, the voltage variance of several voltage fluctuation prediction curves is calculated to obtain a voltage variance fluctuation curve, which reflects the degree of discreteness of the predicted voltage in different time periods. The larger the variance, the worse the consistency of the prediction result, and the first energy storage unit has a balanced abnormality.

[0047] Subsequently, the ratio of the length of the portion of the voltage variance fluctuation curve that is greater than or equal to the balance variance threshold to the total curve length is statistically calculated, and is defined as the voltage balance abnormality coefficient. Among them, the balance variance threshold is a preset parameter used to determine whether the voltage variance is abnormal. If the voltage variance exceeds the threshold, it is considered that there is a balance problem in the first energy storage unit in the corresponding time period. Then, it is determined whether the voltage balance abnormality coefficient is greater than or equal to the voltage balance abnormality coefficient threshold. If so, it means that there is a balance abnormality problem in the first energy storage unit, and fault detection and maintenance are required. At this time, a second energy storage unit fault detection instruction is generated, and a dynamic fault detection process for battery balance problems is started.

[0048] By further detecting the abnormal battery balancing problem of the energy storage unit under the condition that the voltage and temperature fluctuations are normal. Although this abnormality will not immediately lead to obvious performance degradation or safety hazards, its long-term existence will affect the consistency and reliability of the first energy storage unit, and ultimately reduce the service life of the energy storage power station. Therefore, timely detection and processing of battery balancing abnormalities and handling them can provide guarantees for the safe and reliable operation of the energy storage power station.

[0049] Furthermore, the embodiment of the present application also includes:

[0050] S310: The charge and discharge control parameters include charge control parameters or discharge control parameters;

[0051] S320: Predicting battery voltage fluctuation according to the charging control parameter, the ambient temperature monitoring information, and the ambient humidity monitoring information, to obtain a charging voltage fluctuation prediction curve;

[0052] S330: Predicting battery voltage fluctuation according to the discharge control parameter, the ambient temperature monitoring information, and the ambient humidity monitoring information, to obtain a discharge voltage fluctuation prediction curve;

[0053] S340: Setting the charging voltage fluctuation prediction curve or the discharging voltage fluctuation prediction curve as the voltage fluctuation prediction curve.

[0054] In a preferred embodiment, two different working states of charging and discharging in the first energy storage unit are considered respectively, and voltage fluctuation prediction is performed respectively to improve the pertinence and accuracy of the prediction.

[0055] First, the type of charge and discharge control parameters is clarified, that is, charge control parameters or discharge control parameters. Charge control parameters are a series of control indicators used in the charging process of the energy storage unit, such as charge current, charge voltage, charge time, etc., which determine the rate, duration and termination conditions of the charging process; discharge control parameters are a series of control indicators used in the discharge process of the energy storage unit, such as discharge current, discharge voltage, discharge time, etc., which determine the intensity, duration and cut-off conditions of the discharge process. Then, the battery voltage fluctuation is predicted for the charging process and the discharge process respectively. Among them, according to the charge control parameters, the ambient temperature monitoring information and the ambient humidity monitoring information, a battery voltage prediction model under charging state is established, and the charge control parameters are used as input, and the influence of ambient temperature and humidity are considered at the same time, and the prediction curve of the battery voltage change over time during the charging process, that is, the charge voltage fluctuation prediction curve, is output. Similarly, according to the discharge control parameters, the ambient temperature monitoring information and the ambient humidity monitoring information, a battery voltage prediction model under discharge state is established, and the discharge control parameters are used as input, and the influence of ambient temperature and humidity are considered at the same time, and the prediction curve of the battery voltage change over time during the discharge process, that is, the discharge voltage fluctuation prediction curve, is output. After obtaining the charging voltage fluctuation prediction curve and the discharging voltage fluctuation prediction curve, one of them is selected as the voltage fluctuation prediction curve according to the actual working state of the energy storage unit. Specifically, if the first energy storage unit is currently in a charging state, the charging voltage fluctuation prediction curve is used as the voltage fluctuation prediction curve; if the first energy storage unit is currently in a discharging state, the discharging voltage fluctuation prediction curve is used as the voltage fluctuation prediction curve.

[0056] By generating corresponding voltage fluctuation prediction curves for the charging and discharging processes of energy storage units, the particularity of the charging and discharging process is fully considered, and the dynamic change law of battery voltage can be more accurately described. Compared with simply using a single prediction model, while improving the prediction accuracy, it also enhances the reliability and applicability of fault detection, can better cope with the actual working conditions of energy storage power stations, adapt to different charging and discharging modes and environmental conditions, and provide more accurate and comprehensive data support for the health monitoring and fault diagnosis of energy storage units.

[0057] Furthermore, the embodiment of the present application also includes:

[0058] S310: performing neighborhood hierarchical aggregation on the ambient temperature monitoring information and the ambient humidity monitoring information to obtain the first time zone ambient temperature monitoring information and the first time zone humidity monitoring information, until the Nth time zone ambient temperature monitoring information and the Nth time zone humidity monitoring information;

[0059] S320: Segment the charging control parameters according to the first time zone ambient temperature monitoring information and the first time zone humidity monitoring information, up to the Nth time zone ambient temperature monitoring information and the Nth time zone humidity monitoring information, to obtain the first time zone charging control parameters up to the Nth time zone charging control parameters;

[0060] S330: Predicting battery voltage fluctuation according to the charging control parameter in the first time zone, the ambient temperature monitoring information in the first time zone, the humidity monitoring information in the first time zone, and the initial voltage of the energy storage battery, to obtain a charging voltage fluctuation prediction curve in the first time zone;

[0061] S340: until a battery voltage fluctuation prediction is performed according to the charging control parameter of the Nth time zone, the ambient temperature monitoring information of the Nth time zone, the humidity monitoring information of the Nth time zone and the predicted voltage at the end time of the N+1th time zone, to obtain a charging voltage fluctuation prediction curve of the Nth time zone;

[0062] S350: sequentially connecting the first time zone charging voltage fluctuation prediction curves end to end until the Nth time zone charging voltage fluctuation prediction curve to obtain the charging voltage fluctuation prediction curve.

[0063] In a preferred embodiment, when obtaining the charging voltage fluctuation prediction curve, the time correlation of the ambient temperature and humidity and the charging control parameters is taken into consideration, and a more refined voltage prediction is achieved by segmenting and aggregating the time dimension.

[0064] First, the ambient temperature monitoring information and the ambient humidity monitoring information are aggregated at the neighborhood level. Specifically, the entire charging process is divided into multiple time zones, and the ambient temperature and humidity in each time zone are aggregated into a representative value to reduce the redundancy and noise of the environmental data and extract the main trend characteristics. After aggregation, the ambient temperature monitoring information and humidity monitoring information in the first time zone correspond to the temperature and humidity levels at the beginning of charging, respectively, and so on, until the Nth time zone corresponds to the end of charging, that is, the ambient temperature monitoring information in the Nth time zone and the humidity monitoring information in the Nth time zone. Through the time stratification method, the dynamic changes of environmental factors during the charging process can be more accurately characterized. Then, according to the aggregated time zone ambient temperature and humidity information, the charging control parameters are divided accordingly, and the charging control parameters are divided into N time zones with the same environmental factors. Each time zone corresponds to a set of independent charge and discharge control parameters, forming the charging control parameters of the first time zone until the charging control parameters of the Nth time zone. Among them, the first time zone charging control parameter represents the charging strategy under the first time zone ambient temperature monitoring information and the first time zone humidity monitoring information, the second time zone charging control parameter represents the charging strategy under the second time zone ambient temperature monitoring information and the second time zone humidity monitoring information, and so on.

[0065] Afterwards, the battery voltage fluctuation of each time zone is predicted respectively. For the first time zone, the input of the battery voltage fluctuation prediction model includes the charging control parameters of the first time zone, the environmental temperature monitoring information of the first time zone, the humidity monitoring information of the first time zone, and the initial voltage of the energy storage battery. Among them, the initial voltage of the energy storage battery refers to the battery voltage value at the beginning of this charging, which reflects the initial state of the battery. The model output is the charging voltage fluctuation prediction curve of the first time zone, which represents the expected change trend of the battery voltage over time in the first time zone. Similarly, for the second time zone to the Nth time zone, the input of the battery voltage fluctuation prediction model includes the charging control parameters of the corresponding time zone, the environmental temperature and humidity information, and the terminal voltage prediction value of the previous time zone. Among them, the terminal voltage prediction value of the previous time zone is used as the initial voltage of the current time zone to ensure the continuity of the prediction curves of different time zones. The model output is the charging voltage fluctuation prediction curve of each time zone. Afterwards, the charging voltage fluctuation prediction curves of each time zone are connected end to end in chronological order to obtain a complete charging voltage fluctuation prediction curve, covering the entire charging process, reflecting the dynamic change law of the battery voltage under different periods, different environments and control conditions.

[0066] The time-stratified voltage prediction scheme can more finely characterize the changing characteristics of battery voltage at different stages, and improve the time resolution and dynamic adaptability of the prediction. At the same time, by aggregating environmental data and segmenting the charge and discharge control parameters, the redundancy and noise of the model input are reduced, and the key influencing factors are highlighted. In addition, time-zone modeling enables timely feedback of prediction results, thereby realizing real-time fault detection.

[0067] Furthermore, the embodiment of the present application also includes:

[0068] S331: performing a first-level healthy sample search for the first energy storage unit model according to the first time zone charging control parameter, the first time zone ambient temperature monitoring information, the first time zone humidity monitoring information and the initial voltage of the energy storage battery, and obtaining at least 500 first-level sample charging voltage fluctuation record curves, wherein any first-level sample charging voltage fluctuation record curve has the first time zone sample charging control parameter, the first time zone sample ambient temperature monitoring information, the first time zone sample humidity monitoring information and the initial voltage of the energy storage battery;

[0069] S332: Perform a second-level healthy sample search on the first energy storage unit model based on the first time zone sample charging control parameter, the first time zone sample ambient temperature monitoring information, the first time zone sample humidity monitoring information and the energy storage battery initial voltage of each first-level sample charging voltage fluctuation record curve, and obtain at least 500 second-level sample charging voltage fluctuation record curves;

[0070] S333: Merge the secondary sample charging voltage fluctuation record curve and the primary sample charging voltage fluctuation record curve to obtain the first time zone charging voltage fluctuation prediction curve.

[0071] In a preferred embodiment, when obtaining the charging voltage fluctuation prediction curve in the first time zone, based on big data mining and hierarchical search, charging curves similar to the current operating conditions are extracted from massive historical samples and fusion analysis is performed to obtain the final prediction result.

[0072] The historical database stores multiple charging voltage fluctuation record curves, each of which has a first time zone sample charging control parameter, a first time zone sample ambient temperature monitoring information, a first time zone sample humidity monitoring information, and an energy storage battery initial voltage. First, the historical database is searched by the model of the first energy storage unit to obtain multiple candidate charging voltage fluctuation record curves. Then, according to the first time zone charging control parameter, the first time zone ambient temperature monitoring information, the first time zone humidity monitoring information, and the energy storage battery initial voltage, a first-level healthy sample search is performed on multiple candidate charging voltage fluctuation record curves to obtain a first-level sample charging voltage fluctuation record curve. The so-called first-level sample refers to a charging voltage fluctuation record curve that is highly matched with the first time zone operating condition parameters of the current first energy storage unit in the historical database. In order to ensure the representativeness and reliability of the sample, the search result is required to contain at least 500 curves. Then, based on the first-level sample charging voltage fluctuation record curve, a second-level healthy sample search is further performed in the historical database. The second-level sample is an extension and supplement to the first-level sample, aiming to dig out a charging voltage fluctuation record curve similar to the first-level sample. Specifically, for each first-level sample charging voltage fluctuation record curve, the system re-searches for samples with similar parameters to the first time zone of the corresponding historical charging process, requiring the sample size to be no less than 500, to obtain the second-level sample charging voltage fluctuation record curve. Through the second-level sample search, while ensuring similarity, more changes and noise are introduced to improve the robustness and generalization ability of the prediction. At the same time, the addition of a large number of second-level samples also helps to reduce the impact of individual abnormal samples and improve the stability of the prediction results. After obtaining the first-level sample charging voltage fluctuation record curve and the second-level sample charging voltage fluctuation record curve, these curves are fused, and the effective information in each sample curve is comprehensively utilized to obtain a prediction result that can represent the overall trend as the first time zone charging voltage fluctuation prediction curve.

[0073] By utilizing the rules of historical operating data, based on similarity search and stratified sampling, the reference samples most relevant to the current scenario are found. On this basis, a fusion strategy is adopted to integrate the sample curves to obtain a reliable and accurate charging voltage fluctuation prediction curve for the first time zone.

[0074] Furthermore, the embodiment of the present application also includes:

[0075] S3331: Obtaining a set of second-level first-sample charging voltage fluctuation record curves of the first-level first-sample charging voltage fluctuation record curve;

[0076] S3332: performing pairwise distance evaluation on the set of second-level first sample charging voltage fluctuation record curves to obtain a set of distribution distance parameters of the second-level first sample charging voltage fluctuation record curves;

[0077] S3333: performing LOF outlier analysis on the set of the second-level first sample charging voltage fluctuation record curves according to the distribution distance parameter set of the second-level first sample charging voltage fluctuation record curves to obtain a set of outlier factors of the second-level first sample charging voltage fluctuation record curves;

[0078] S3334: extracting the minimum value of the outlier factor set of the secondary first sample charging voltage fluctuation record curve from the secondary first sample charging voltage fluctuation record curve set, obtaining the centroid sample charging voltage fluctuation record curve, adding it to the primary sample charging voltage fluctuation record curve, and obtaining the primary sample updated charging voltage fluctuation record curve;

[0079] S3335: When the analysis of the secondary sample charging voltage fluctuation record curve is completed, the centroid curve analysis is performed on the primary sample updated charging voltage fluctuation record curve to obtain the first time zone charging voltage fluctuation prediction curve.

[0080] In a preferred embodiment, when the secondary sample charging voltage fluctuation record curve and the primary sample charging voltage fluctuation record curve are merged, first, the primary sample charging voltage fluctuation record curves are traversed, and each curve is obtained as the primary first sample charging voltage fluctuation record curve, and the secondary first sample charging voltage fluctuation record curve set corresponding to the primary first sample charging voltage fluctuation record curve is extracted. Then, for each pair of curves in the secondary first sample charging voltage fluctuation record curve set, the similarity or distance metric between the two curves is calculated, such as Euclidean distance, Manhattan distance, etc., to obtain the secondary first sample charging voltage fluctuation record curve distribution distance parameter set, quantitatively characterize the similarity between different sample curves, and provide a basis for subsequent outlier analysis and centroid selection.

[0081] Next, the LOF algorithm is used to perform outlier analysis on the distribution distance parameter set of the secondary first sample charging voltage fluctuation record curve according to the distribution distance parameter set of the secondary first sample charging voltage fluctuation record curve. The LOF algorithm is a density-based local anomaly detection method, which judges the degree of outlier of the sample point by comparing the relative density of the sample point with its neighborhood point. Specifically, for each secondary first sample charging voltage fluctuation record curve, its local neighborhood range is determined according to the distance parameter obtained from the distribution distance parameter of the secondary first sample charging voltage fluctuation record curve; then, the average relative density of the curve and other curves in the neighborhood is calculated to obtain its LOF value as the outlier factor, and the outlier factor set of the secondary first sample charging voltage fluctuation record curve is obtained to quantitatively evaluate the degree of abnormality of each sample curve, and provide a quantitative indicator for the selection of the centroid curve. The larger the LOF value, the greater the relative density difference between the curve and the surrounding curves, that is, the higher the degree of outlier.

[0082] Subsequently, from the set of charging voltage fluctuation record curves of the second-level first sample, the curve with the smallest outlier factor is extracted as the centroid sample according to the outlier factor set of the charging voltage fluctuation record curves of the second-level first sample as the centroid sample, that is, the charging voltage fluctuation record curve of the centroid sample. Among them, the smallest outlier factor means that the relative density difference between the curve and other curves in its neighborhood is the smallest, that is, the most representative and central. Selecting such a curve as the centroid can ensure the robustness and accuracy of the fusion result. The charging voltage fluctuation record curve of the centroid sample is added to the charging voltage fluctuation record curve of the first-level sample to realize the information transmission and update from the second-level sample to the first-level sample, enriching the diversity and representativeness of the first-level sample. Repeat the execution until all the charging voltage fluctuation record curves of the second-level samples are analyzed and processed. In this process, the charging voltage fluctuation record curve of the first-level sample is continuously updated and expanded, and the centroid curves from different charging voltage fluctuation record curves of the second-level samples are incorporated, so as to obtain the updated charging voltage fluctuation record curve of the first-level sample that integrates the characteristics of each second-level sample. Afterwards, pairwise distance evaluation and LOF outlier analysis are performed on the first-level sample updated charging voltage fluctuation record curve again to realize the centroid curve analysis, so as to select the most representative and central curve from the first-level sample updated charging voltage fluctuation record curve as the charging voltage fluctuation prediction curve for the first time zone.

[0083] Furthermore, the embodiment of the present application also includes:

[0084] S3321: According to the first energy storage unit model, obtain the charging control parameters of the healthy sample to be selected, the environmental temperature monitoring information of the healthy sample to be selected, the humidity monitoring information of the healthy sample to be selected, and the initial voltage of the energy storage battery of the healthy sample to be selected;

[0085] S3322: When the charging control parameter distance between the charging control parameter of the candidate healthy sample and the charging control parameter of the first time zone is less than or equal to the charging control parameter distance threshold, it is deemed that the charging control parameter of the candidate healthy sample meets the first sorting condition;

[0086] S3323: When the ambient temperature deviation between the ambient temperature monitoring information of the selected healthy sample and the ambient temperature monitoring information of the first time zone is less than or equal to the ambient temperature deviation threshold, it is considered that the charging control parameter of the selected healthy sample meets the second sorting condition;

[0087] S3324: When the humidity deviation between the humidity monitoring information of the selected healthy sample and the humidity monitoring information of the first time zone is less than or equal to the ambient humidity deviation threshold, it is considered that the charging control parameter of the selected healthy sample meets the third sorting condition;

[0088] S3325: When the initial voltage of the energy storage battery of the to-be-selected healthy sample is the same as the initial voltage of the energy storage battery, it is deemed that the charging control parameter of the to-be-selected healthy sample meets the fourth sorting condition;

[0089] S3326: When the first sorting condition, the second sorting condition, the third sorting condition and the fourth sorting condition are all met, the charging voltage fluctuation record curve of the healthy sample to be selected is added to the charging voltage fluctuation record curve of the first-level sample.

[0090] In a preferred embodiment, to realize the primary healthy sample search, first, according to the first energy storage unit model, the relevant parameters of the charging voltage fluctuation record curve of the selected healthy sample are obtained, including the charging control parameters of the selected healthy sample, the environmental temperature monitoring information of the selected healthy sample, the humidity monitoring information of the selected healthy sample, and the initial voltage of the energy storage battery of the selected healthy sample. These parameters, together with the corresponding parameters of the first time zone, constitute the comparison basis for sample screening.

[0091] Then, the distance between the charging control parameter of the selected healthy sample and the charging control parameter of the first time zone is calculated, and compared with the preset charging control parameter distance threshold. If the distance is less than or equal to the threshold, it is considered that the charging voltage fluctuation record curve of the selected healthy sample is highly similar to the current first time zone in terms of charging control strategy and working conditions, and meets the first sorting condition. At the same time, the deviation between the environmental temperature monitoring information of the selected healthy sample and the environmental temperature monitoring information of the first time zone is calculated, and compared with the preset environmental temperature deviation threshold. If the deviation is less than or equal to the threshold, it is considered that the charging voltage fluctuation record curve of the selected healthy sample is highly similar to the current first time zone in terms of environmental temperature conditions, and meets the second sorting condition. At the same time, the deviation between the humidity monitoring information of the selected healthy sample and the humidity monitoring information of the first time zone is calculated, and compared with the preset environmental humidity deviation threshold. If the deviation is less than or equal to the threshold, it is considered that the charging voltage fluctuation record curve of the selected healthy sample is highly similar to the current first time zone in terms of environmental humidity conditions, and meets the third sorting condition. At the same time, compare whether the initial voltage of the energy storage battery of the selected healthy sample is the same as the initial voltage of the energy storage battery in the first time zone. If the two are exactly the same, it is considered that the charging voltage fluctuation record curve of the candidate healthy sample completely matches the current first time zone in terms of the battery state before charging, and meets the fourth sorting condition. Only when a candidate healthy sample charging voltage fluctuation record curve meets the first to fourth sorting conditions at the same time can it be regarded as a first-level healthy sample that is highly matched with the current first time zone charging control parameters, providing reliable and effective support for subsequent voltage fluctuation predictions. The candidate sample charging voltage fluctuation record curve that meets all sorting conditions is added to the first-level sample charging voltage fluctuation record curve, and the number of first-level sample charging voltage fluctuation record curves is required to be no less than 500 to ensure the richness and representativeness of the sample.

[0092] Furthermore, the embodiment of the present application also includes:

[0093] S33221: Construct charging control parameter distance evaluation function:

[0094]

[0095] Among them, SIM i The distance that characterizes the charging control parameter of the i-th dimension, T 1i Characterizes the end time of the charging control parameters in the first time zone, T 2i Characterizes the end time of the charging control parameters of the sample to be sorted, x 1ij The j-th time characteristic value of the i-th dimension charging control parameter representing the charging control parameter of the first time zone, x 2ij The characteristic value of the i-th dimension charging control parameter at the jth moment that represents the charging control parameter of the sample to be sorted, b is a small constant, a is the adjustment coefficient, 1>a>0;

[0096] S33222: The charging control parameter distance threshold includes charging control parameter distance thresholds of several dimensions;

[0097] S33223: Analyze the charging control parameter distances of several dimensions between the charging control parameter of the selected healthy sample and the charging control parameter of the first time zone according to the charging control parameter distance evaluation function;

[0098] S33224: When the distances of the charging control parameters of the multiple dimensions in each dimension meet the distance thresholds of the charging control parameters of the multiple dimensions, it is considered that the charging control parameters of the healthy sample to be selected meet the first sorting condition;

[0099] S33225: Otherwise, it is deemed that the charging control parameters of the healthy sample to be selected do not meet the first sorting condition.

[0100] In a preferred embodiment, in order to determine whether the charging control parameters of the selected healthy samples meet the first sorting condition, firstly, a charging control parameter distance evaluation function is constructed as follows:

[0101]

[0102] Among them, |T 1i -T 2i | represents the absolute value of the end point time difference between the charging control parameter of the first time zone and the charging control parameter of the sample to be sorted in the i-th dimension. The larger the end point time difference, the greater the difference in charging duration of the two samples in this dimension, and the greater the distance value. represents the cosine similarity of the eigenvalues ​​of two samples in the common time period of the i-th dimension charging control parameter. 1ijand x 2ij They represent the characteristic values ​​of the charging control parameters of the first time zone and the charging control parameters of the sample to be sorted at the jth moment in the i-th dimension. The larger the cosine similarity, the closer the change trends of the two samples in this dimension are, and the smaller the distance value is. a is a logarithmic function, a is an adjustment coefficient, and its value is between 0 and 1. The purpose of introducing the logarithmic function is to scale the distance value so that the distance values ​​of different dimensions are closer in magnitude, which is convenient for subsequent comprehensive comparison. The smaller the a value, the more obvious the compression effect of the logarithmic function. b is a small constant used to avoid the situation where the denominator is zero.

[0103] At the same time, the charging control parameter distance threshold includes several dimensional charging control parameter distance thresholds, and each dimension corresponds to a threshold. Then, the charging control parameter distance evaluation function is used to calculate the distance values ​​between the charging control parameters of the selected healthy samples and the charging control parameters of the first time zone in each dimension, and several dimensional charging control parameter distances are obtained. After that, it is determined whether the charging control parameter distances of each dimension meet the corresponding charging control parameter distance threshold. If the charging control parameter distances of all dimensions are less than or equal to the corresponding charging control parameter distance threshold, it is considered that the charging control parameters of the selected healthy samples are close enough to the first time zone in terms of charging control parameters, and the first sorting condition is met. If there is any dimension whose charging control parameter distance exceeds the corresponding charging control parameter distance threshold, it is considered that the charging control parameters of the selected healthy samples are greatly different from those of the first time zone in terms of charging control parameters, and the first sorting condition is not met.

[0104] By introducing a multi-dimensional distance evaluation function, the characteristics of different charging control parameters are comprehensively considered to improve the accuracy of screening. At the same time, by adjusting the parameters of the distance function, such as changing the a value and b value, the distribution and sensitivity of the distance value can be flexibly controlled to meet different application requirements.

[0105] In summary, the fault detection method of an energy storage power station provided in the embodiment of the present application has the following technical effects:

[0106] Obtain the charge and discharge control parameters of the first energy storage unit of the energy storage power station to provide basic data support for subsequent fault detection. Interact with environmental sensors to receive ambient temperature monitoring information and ambient humidity monitoring information to provide information for obtaining voltage fluctuation prediction curves and temperature fluctuation prediction curves. Predict battery voltage fluctuations based on charge and discharge control parameters, ambient temperature monitoring information, and ambient humidity monitoring information to obtain voltage fluctuation prediction curves, grasp the voltage change law of the battery under normal working conditions, and provide a reference for subsequent fault diagnosis. Predict battery temperature fluctuations based on charge and discharge control parameters, ambient temperature monitoring information, and ambient humidity monitoring information to obtain temperature fluctuation prediction curves, grasp the temperature change law of the battery under normal working conditions, and provide a reference for subsequent fault diagnosis. Obtain voltage fluctuation monitoring curves and temperature fluctuation monitoring curves, obtain voltage and temperature change data during actual operation, and provide a judgment basis for subsequent fault diagnosis. Compare the voltage fluctuation prediction curve with the voltage fluctuation monitoring curve to obtain the voltage fluctuation deviation coefficient, and compare the temperature fluctuation prediction curve with the temperature fluctuation monitoring curve to obtain the temperature fluctuation deviation coefficient, wherein the voltage fluctuation deviation coefficient represents the proportion of moments greater than or equal to the voltage deviation threshold, and the temperature fluctuation deviation coefficient represents the proportion of moments greater than or equal to the temperature deviation threshold, quantitatively evaluating the degree of difference between the actual operating state of the battery and the expected state. When the voltage fluctuation deviation coefficient is greater than or equal to the voltage fluctuation deviation coefficient threshold, or / and the temperature fluctuation deviation coefficient is greater than or equal to the temperature fluctuation deviation coefficient threshold, a first energy storage unit fault detection instruction is generated to perform dynamic fault detection, achieve early warning and accurate diagnosis of battery faults, reduce the operation and maintenance costs of energy storage power stations, and improve their safety and reliability.

[0107] Embodiment 2 is based on the same inventive concept as the fault detection method of an energy storage power station in the above embodiment. Figure 2 As shown, an embodiment of the present application provides a fault detection device for an energy storage power station, the device comprising:

[0108] A parameter acquisition module 11 is used to obtain the charge and discharge control parameters of the first energy storage unit of the energy storage power station;

[0109] The environment monitoring module 12 is used to interact with the environment sensor and receive the environment temperature monitoring information and the environment humidity monitoring information;

[0110] A voltage prediction module 13, configured to predict battery voltage fluctuations according to the charge and discharge control parameters, the ambient temperature monitoring information, and the ambient humidity monitoring information, and obtain a voltage fluctuation prediction curve;

[0111] A temperature prediction module 14, configured to predict battery temperature fluctuations according to the charge and discharge control parameters, the ambient temperature monitoring information, and the ambient humidity monitoring information, and obtain a temperature fluctuation prediction curve;

[0112] The measured data acquisition module 15 is used to obtain the voltage fluctuation monitoring curve and the temperature fluctuation monitoring curve;

[0113] The deviation analysis module 16 is used to compare the voltage fluctuation prediction curve with the voltage fluctuation monitoring curve to obtain a voltage fluctuation deviation coefficient, and compare the temperature fluctuation prediction curve with the temperature fluctuation monitoring curve to obtain a temperature fluctuation deviation coefficient, wherein the voltage fluctuation deviation coefficient represents a proportion of moments greater than or equal to a voltage deviation threshold, and the temperature fluctuation deviation coefficient represents a proportion of moments greater than or equal to a temperature deviation threshold;

[0114] The fault detection module 17 is used to generate a first energy storage unit fault detection instruction to perform dynamic fault detection when the voltage fluctuation deviation coefficient is greater than or equal to the voltage fluctuation deviation coefficient threshold, or / and the temperature fluctuation deviation coefficient is greater than or equal to the temperature fluctuation deviation coefficient threshold.

[0115] Furthermore, the embodiment of the present application also includes:

[0116] When the voltage fluctuation deviation coefficient is less than the voltage fluctuation deviation coefficient threshold, and the temperature fluctuation deviation coefficient is less than the temperature fluctuation deviation coefficient threshold, obtaining voltage variance fluctuation curves of a plurality of voltage fluctuation prediction curves of the first energy storage unit;

[0117] Counting the proportion of the curve length greater than or equal to the balance variance threshold in the voltage variance fluctuation curve, and setting it as the voltage balance abnormality coefficient;

[0118] When the voltage balance abnormality coefficient is greater than or equal to the voltage balance abnormality coefficient threshold, a second energy storage unit fault detection instruction is generated to perform dynamic fault detection.

[0119] Furthermore, the voltage prediction module 13 includes the following execution steps:

[0120] The charge and discharge control parameters include charge control parameters or discharge control parameters;

[0121] Predicting battery voltage fluctuation according to the charging control parameter, the ambient temperature monitoring information, and the ambient humidity monitoring information to obtain a charging voltage fluctuation prediction curve;

[0122] Perform battery voltage fluctuation prediction according to the discharge control parameter, the ambient temperature monitoring information, and the ambient humidity monitoring information to obtain a discharge voltage fluctuation prediction curve;

[0123] The charging voltage fluctuation prediction curve or the discharging voltage fluctuation prediction curve is set as the voltage fluctuation prediction curve.

[0124] Furthermore, the voltage prediction module 13 further includes the following execution steps:

[0125] Performing neighborhood hierarchical aggregation on the ambient temperature monitoring information and the ambient humidity monitoring information to obtain the first time zone ambient temperature monitoring information and the first time zone humidity monitoring information, until the Nth time zone ambient temperature monitoring information and the Nth time zone humidity monitoring information;

[0126] According to the first time zone ambient temperature monitoring information and the first time zone humidity monitoring information, up to the Nth time zone ambient temperature monitoring information and the Nth time zone humidity monitoring information, the charging control parameter is divided to obtain the first time zone charging control parameter up to the Nth time zone charging control parameter;

[0127] Predicting battery voltage fluctuations according to the charging control parameter in the first time zone, the ambient temperature monitoring information in the first time zone, the humidity monitoring information in the first time zone, and the initial voltage of the energy storage battery to obtain a charging voltage fluctuation prediction curve in the first time zone;

[0128] Until the battery voltage fluctuation is predicted according to the charging control parameter of the Nth time zone, the ambient temperature monitoring information of the Nth time zone, the humidity monitoring information of the Nth time zone and the predicted voltage at the end time of the N+1th time zone, to obtain a charging voltage fluctuation prediction curve of the Nth time zone;

[0129] The charging voltage fluctuation prediction curves in the first time zones are sequentially connected end to end until the charging voltage fluctuation prediction curve in the Nth time zone to obtain the charging voltage fluctuation prediction curve.

[0130] Furthermore, the voltage prediction module 13 further includes the following execution steps:

[0131] Performing a first-level healthy sample search for the first energy storage unit model according to the first time zone charging control parameter, the first time zone ambient temperature monitoring information, the first time zone humidity monitoring information, and the initial voltage of the energy storage battery, and obtaining at least 500 first-level sample charging voltage fluctuation record curves, wherein any first-level sample charging voltage fluctuation record curve has the first time zone sample charging control parameter, the first time zone sample ambient temperature monitoring information, the first time zone sample humidity monitoring information, and the initial voltage of the energy storage battery;

[0132] Perform a secondary health sample search on the first energy storage unit model for each first-level sample charging voltage fluctuation record curve based on the first time zone sample charging control parameter, the first time zone sample ambient temperature monitoring information, the first time zone sample humidity monitoring information, and the energy storage battery initial voltage, to obtain at least 500 second-level sample charging voltage fluctuation record curves;

[0133] The secondary sample charging voltage fluctuation record curve and the primary sample charging voltage fluctuation record curve are merged to obtain the first time zone charging voltage fluctuation prediction curve.

[0134] Furthermore, the voltage prediction module 13 further includes the following execution steps:

[0135] Obtaining a set of second-level first-sample charging voltage fluctuation record curves of the first-level first-sample charging voltage fluctuation record curve;

[0136] Performing pairwise distance evaluation on the secondary first sample charging voltage fluctuation record curve set to obtain a secondary first sample charging voltage fluctuation record curve distribution distance parameter set;

[0137] According to the distribution distance parameter set of the secondary first sample charging voltage fluctuation record curve, performing LOF outlier analysis on the secondary first sample charging voltage fluctuation record curve set to obtain an outlier factor set of the secondary first sample charging voltage fluctuation record curve;

[0138] Extracting the minimum value of the outlier factor set of the secondary first sample charging voltage fluctuation record curve from the secondary first sample charging voltage fluctuation record curve set, obtaining the centroid sample charging voltage fluctuation record curve, adding it to the primary sample charging voltage fluctuation record curve, and obtaining the primary sample updated charging voltage fluctuation record curve;

[0139] When all the analysis of the secondary sample charging voltage fluctuation record curves is completed, the centroid curve analysis is performed on the primary sample updated charging voltage fluctuation record curve to obtain the first time zone charging voltage fluctuation prediction curve.

[0140] Furthermore, the voltage prediction module 13 further includes the following execution steps:

[0141] According to the first energy storage unit model, obtaining the charging control parameters of the healthy sample to be selected, the environmental temperature monitoring information of the healthy sample to be selected, the humidity monitoring information of the healthy sample to be selected, and the initial voltage of the energy storage battery of the healthy sample to be selected;

[0142] When the charging control parameter distance between the charging control parameter of the to-be-selected healthy sample and the charging control parameter of the first time zone is less than or equal to the charging control parameter distance threshold, it is deemed that the charging control parameter of the to-be-selected healthy sample meets the first sorting condition;

[0143] When the ambient temperature deviation between the ambient temperature monitoring information of the selected healthy sample and the ambient temperature monitoring information of the first time zone is less than or equal to the ambient temperature deviation threshold, it is considered that the charging control parameter of the selected healthy sample meets the second sorting condition;

[0144] When the humidity deviation between the humidity monitoring information of the selected healthy sample and the humidity monitoring information of the first time zone is less than or equal to the ambient humidity deviation threshold, it is considered that the charging control parameter of the selected healthy sample meets the third sorting condition;

[0145] When the initial voltage of the energy storage battery of the to-be-selected healthy sample is the same as the initial voltage of the energy storage battery, it is deemed that the charging control parameter of the to-be-selected healthy sample meets the fourth sorting condition;

[0146] When the first sorting condition, the second sorting condition, the third sorting condition and the fourth sorting condition are all met, the charging voltage fluctuation record curve of the to-be-selected healthy sample is added to the charging voltage fluctuation record curve of the first-level sample.

[0147] Furthermore, the voltage prediction module 13 further includes the following execution steps:

[0148] Construct the charging control parameter distance evaluation function:

[0149]

[0150] Among them, SIM i The distance that characterizes the charging control parameter of the i-th dimension, T 1i Characterizes the end time of the charging control parameters in the first time zone, T 2i Characterizes the end time of the charging control parameters of the sample to be sorted, x 1ij The j-th time characteristic value of the i-th dimension charging control parameter representing the charging control parameter of the first time zone, x 2ij The characteristic value of the i-th dimension charging control parameter at the jth moment that represents the charging control parameter of the sample to be sorted, b is a small constant, a is the adjustment coefficient, 1>a>0;

[0151] The charging control parameter distance threshold includes several dimensional charging control parameter distance thresholds;

[0152] According to the charging control parameter distance evaluation function, analyzing the charging control parameter distances of several dimensions between the charging control parameter of the selected healthy sample and the charging control parameter of the first time zone;

[0153] When the distances of the charging control parameters of the plurality of dimensions in each dimension meet the distance thresholds of the charging control parameters of the plurality of dimensions, it is considered that the charging control parameters of the healthy sample to be selected meet the first sorting condition;

[0154] Otherwise, it is regarded that the charging control parameter of the to-be-selected healthy sample does not meet the first sorting condition.

[0155] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0156] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. 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 equivalents, the present application intends to include these changes and variations.

Claims

1. A method for detecting a fault in an energy storage power station, characterized in that: include: Obtaining charging and discharging control parameters of a first energy storage unit of an energy storage power station; Interact with environmental sensors to receive environmental temperature monitoring information and environmental humidity monitoring information; Predicting battery voltage fluctuations according to the charge and discharge control parameters, the ambient temperature monitoring information, and the ambient humidity monitoring information to obtain a voltage fluctuation prediction curve; Predicting battery temperature fluctuations according to the charge and discharge control parameters, the ambient temperature monitoring information, and the ambient humidity monitoring information to obtain a temperature fluctuation prediction curve; Obtain voltage fluctuation monitoring curve and temperature fluctuation monitoring curve; Compare the voltage fluctuation prediction curve with the voltage fluctuation monitoring curve to obtain a voltage fluctuation deviation coefficient, and compare the temperature fluctuation prediction curve with the temperature fluctuation monitoring curve to obtain a temperature fluctuation deviation coefficient, wherein the voltage fluctuation deviation coefficient represents the proportion of moments greater than or equal to a voltage deviation threshold, and the temperature fluctuation deviation coefficient represents the proportion of moments greater than or equal to a temperature deviation threshold; When the voltage fluctuation deviation coefficient is greater than or equal to the voltage fluctuation deviation coefficient threshold, or / and the temperature fluctuation deviation coefficient is greater than or equal to the temperature fluctuation deviation coefficient threshold, a first energy storage unit fault detection instruction is generated to perform dynamic fault detection.

2. The method according to claim 1, characterized in that Also includes: When the voltage fluctuation deviation coefficient is less than the voltage fluctuation deviation coefficient threshold, and the temperature fluctuation deviation coefficient is less than the temperature fluctuation deviation coefficient threshold, obtaining voltage variance fluctuation curves of a plurality of voltage fluctuation prediction curves of the first energy storage unit; Counting the proportion of the curve length greater than or equal to the balance variance threshold in the voltage variance fluctuation curve, and setting it as the voltage balance abnormality coefficient; When the voltage balance abnormality coefficient is greater than or equal to the voltage balance abnormality coefficient threshold, a second energy storage unit fault detection instruction is generated to perform dynamic fault detection.

3. The method according to claim 1, characterized in that The battery voltage fluctuation is predicted according to the charge and discharge control parameter, the ambient temperature monitoring information and the ambient humidity monitoring information to obtain a voltage fluctuation prediction curve, including: The charge and discharge control parameters include charge control parameters or discharge control parameters; Predicting battery voltage fluctuation according to the charging control parameter, the ambient temperature monitoring information, and the ambient humidity monitoring information to obtain a charging voltage fluctuation prediction curve; Perform battery voltage fluctuation prediction according to the discharge control parameter, the ambient temperature monitoring information, and the ambient humidity monitoring information to obtain a discharge voltage fluctuation prediction curve; The charging voltage fluctuation prediction curve or the discharging voltage fluctuation prediction curve is set as the voltage fluctuation prediction curve.

4. The method according to claim 3, characterized in that The battery voltage fluctuation prediction is performed according to the charging control parameter, the ambient temperature monitoring information and the ambient humidity monitoring information to obtain a charging voltage fluctuation prediction curve, including: Performing neighborhood hierarchical aggregation on the ambient temperature monitoring information and the ambient humidity monitoring information to obtain the first time zone ambient temperature monitoring information and the first time zone humidity monitoring information, until the Nth time zone ambient temperature monitoring information and the Nth time zone humidity monitoring information; According to the first time zone ambient temperature monitoring information and the first time zone humidity monitoring information, up to the Nth time zone ambient temperature monitoring information and the Nth time zone humidity monitoring information, the charging control parameter is divided to obtain the first time zone charging control parameter up to the Nth time zone charging control parameter; Predicting battery voltage fluctuations according to the charging control parameter in the first time zone, the ambient temperature monitoring information in the first time zone, the humidity monitoring information in the first time zone, and the initial voltage of the energy storage battery to obtain a charging voltage fluctuation prediction curve in the first time zone; Until the battery voltage fluctuation is predicted according to the charging control parameter of the Nth time zone, the ambient temperature monitoring information of the Nth time zone, the humidity monitoring information of the Nth time zone and the predicted voltage at the end time of the N+1th time zone, to obtain a charging voltage fluctuation prediction curve of the Nth time zone; The charging voltage fluctuation prediction curves in the first time zones are sequentially connected end to end until the charging voltage fluctuation prediction curve in the Nth time zone to obtain the charging voltage fluctuation prediction curve.

5. The method according to claim 4, characterized in that The battery voltage fluctuation prediction is performed according to the charging control parameter in the first time zone, the ambient temperature monitoring information in the first time zone, the humidity monitoring information in the first time zone, and the initial voltage of the energy storage battery to obtain a charging voltage fluctuation prediction curve in the first time zone, including: Performing a first-level healthy sample search for the first energy storage unit model according to the first time zone charging control parameter, the first time zone ambient temperature monitoring information, the first time zone humidity monitoring information, and the initial voltage of the energy storage battery, and obtaining at least 500 first-level sample charging voltage fluctuation record curves, wherein any first-level sample charging voltage fluctuation record curve has the first time zone sample charging control parameter, the first time zone sample ambient temperature monitoring information, the first time zone sample humidity monitoring information, and the initial voltage of the energy storage battery; Perform a secondary health sample search on the first energy storage unit model for each first-level sample charging voltage fluctuation record curve based on the first time zone sample charging control parameter, the first time zone sample ambient temperature monitoring information, the first time zone sample humidity monitoring information, and the energy storage battery initial voltage, to obtain at least 500 second-level sample charging voltage fluctuation record curves; The secondary sample charging voltage fluctuation record curve and the primary sample charging voltage fluctuation record curve are merged to obtain the first time zone charging voltage fluctuation prediction curve.

6. The method according to claim 5, characterized in that The second-level sample charging voltage fluctuation record curve and the first-level sample charging voltage fluctuation record curve are merged to obtain the first time zone charging voltage fluctuation prediction curve, including: Obtaining a set of second-level first-sample charging voltage fluctuation record curves of the first-level first-sample charging voltage fluctuation record curve; Performing pairwise distance evaluation on the secondary first sample charging voltage fluctuation record curve set to obtain a secondary first sample charging voltage fluctuation record curve distribution distance parameter set; According to the distribution distance parameter set of the secondary first sample charging voltage fluctuation record curve, performing LOF outlier analysis on the secondary first sample charging voltage fluctuation record curve set to obtain an outlier factor set of the secondary first sample charging voltage fluctuation record curve; Extracting the minimum value of the outlier factor set of the secondary first sample charging voltage fluctuation record curve from the secondary first sample charging voltage fluctuation record curve set, obtaining the centroid sample charging voltage fluctuation record curve, adding it to the primary sample charging voltage fluctuation record curve, and obtaining the primary sample updated charging voltage fluctuation record curve; When all the analysis of the secondary sample charging voltage fluctuation record curves is completed, the centroid curve analysis is performed on the primary sample updated charging voltage fluctuation record curve to obtain the first time zone charging voltage fluctuation prediction curve.

7. The method according to claim 5, characterized in that A first-level healthy sample search is performed on the first energy storage unit model according to the first time zone charging control parameter, the first time zone ambient temperature monitoring information, the first time zone humidity monitoring information and the energy storage battery initial voltage, and at least 500 first-level sample charging voltage fluctuation record curves are obtained, including: According to the first energy storage unit model, obtaining the charging control parameters of the healthy sample to be selected, the environmental temperature monitoring information of the healthy sample to be selected, the humidity monitoring information of the healthy sample to be selected, and the initial voltage of the energy storage battery of the healthy sample to be selected; When the charging control parameter distance between the charging control parameter of the to-be-selected healthy sample and the charging control parameter of the first time zone is less than or equal to the charging control parameter distance threshold, it is deemed that the charging control parameter of the to-be-selected healthy sample meets the first sorting condition; When the ambient temperature deviation between the ambient temperature monitoring information of the selected healthy sample and the ambient temperature monitoring information of the first time zone is less than or equal to the ambient temperature deviation threshold, it is considered that the charging control parameter of the selected healthy sample meets the second sorting condition; When the humidity deviation between the humidity monitoring information of the selected healthy sample and the humidity monitoring information of the first time zone is less than or equal to the ambient humidity deviation threshold, it is considered that the charging control parameter of the selected healthy sample meets the third sorting condition; When the initial voltage of the energy storage battery of the to-be-selected healthy sample is the same as the initial voltage of the energy storage battery, it is deemed that the charging control parameter of the to-be-selected healthy sample meets the fourth sorting condition; When the first sorting condition, the second sorting condition, the third sorting condition and the fourth sorting condition are all met, the charging voltage fluctuation record curve of the to-be-selected healthy sample is added to the charging voltage fluctuation record curve of the first-level sample.

8. The method according to claim 7, characterized in that When the charging control parameter distance between the charging control parameter of the to-be-selected healthy sample and the charging control parameter of the first time zone is less than or equal to the charging control parameter distance threshold, it is considered that the charging control parameter of the to-be-selected healthy sample meets the first sorting condition, including: Construct the charging control parameter distance evaluation function: Among them, SIM i The distance that characterizes the charging control parameter of the i-th dimension, T 1i Characterizes the end time of the charging control parameters in the first time zone, T 2i Characterizes the end time of the charging control parameters of the sample to be sorted, x 1ij The j-th time characteristic value of the i-th dimension charging control parameter representing the charging control parameter of the first time zone, x 2ij The characteristic value of the i-th dimension charging control parameter at the jth moment that represents the charging control parameter of the sample to be sorted, b is a small constant, a is the adjustment coefficient, 1>a>0; The charging control parameter distance threshold includes several dimensional charging control parameter distance thresholds; According to the charging control parameter distance evaluation function, analyzing the charging control parameter distances of several dimensions between the charging control parameter of the selected healthy sample and the charging control parameter of the first time zone; When the distances of the charging control parameters of the plurality of dimensions in each dimension meet the distance thresholds of the charging control parameters of the plurality of dimensions, it is considered that the charging control parameters of the healthy sample to be selected meet the first sorting condition; Otherwise, it is regarded that the charging control parameter of the to-be-selected healthy sample does not meet the first sorting condition.

9. A fault detection device for an energy storage power station, characterized in that: A method for detecting a fault in an energy storage power station according to any one of claims 1 to 8, comprising: A parameter acquisition module, the parameter acquisition module is used to obtain the charge and discharge control parameters of the first energy storage unit of the energy storage power station; An environmental monitoring module, the environmental monitoring module is used to interact with environmental sensors and receive environmental temperature monitoring information and environmental humidity monitoring information; A voltage prediction module, the voltage prediction module is used to predict battery voltage fluctuations according to the charge and discharge control parameters, the ambient temperature monitoring information and the ambient humidity monitoring information, and obtain a voltage fluctuation prediction curve; A temperature prediction module, the temperature prediction module is used to predict battery temperature fluctuations according to the charge and discharge control parameters, the ambient temperature monitoring information and the ambient humidity monitoring information, and obtain a temperature fluctuation prediction curve; A measured data acquisition module, wherein the measured data acquisition module is used to obtain a voltage fluctuation monitoring curve and a temperature fluctuation monitoring curve; A deviation analysis module, the deviation analysis module is used to compare the voltage fluctuation prediction curve with the voltage fluctuation monitoring curve to obtain a voltage fluctuation deviation coefficient, and compare the temperature fluctuation prediction curve with the temperature fluctuation monitoring curve to obtain a temperature fluctuation deviation coefficient, wherein the voltage fluctuation deviation coefficient represents a proportion of moments greater than or equal to a voltage deviation threshold, and the temperature fluctuation deviation coefficient represents a proportion of moments greater than or equal to a temperature deviation threshold; A fault detection module is used to generate a first energy storage unit fault detection instruction to perform dynamic fault detection when the voltage fluctuation deviation coefficient is greater than or equal to the voltage fluctuation deviation coefficient threshold, or / and the temperature fluctuation deviation coefficient is greater than or equal to the temperature fluctuation deviation coefficient threshold.

Images