An electrochemical energy storage power plant field monitoring method, system, device and medium

Abstract

The application relates to an electrochemical energy storage power station field monitoring method, which comprises the following steps: obtaining historical operation parameters of each energy storage unit in a target chemical energy storage power station and placing the historical operation parameters into a historical operation data set; performing clustering division to obtain K clustering results, determining the states of the energy storage units corresponding to the clustering results based on the data characteristics of the clustering results; forming a first training sample set and a second training sample set based on the clustering results; training a first neural network through the first training sample set to obtain a first monitoring model for monitoring whether the energy storage unit is healthy; training a second neural network through the second training sample set to obtain a second monitoring model for identifying which abnormal sub-state the energy storage unit belongs to; inputting real-time operation parameters of each energy storage unit into the first monitoring model to determine whether the energy storage unit is healthy; when the energy storage unit is in an unhealthy state, inputting corresponding data of the energy storage unit into the second monitoring model to identify which abnormal sub-state the energy storage unit belongs to.

Description

Technical Field

[0001] This invention relates to a method, system, equipment, and medium for on-site monitoring of electrochemical energy storage power stations, belonging to the field of monitoring technology for electrochemical energy storage power stations. Background Technology

[0002] Electrochemical energy storage power stations are widely distributed and play an increasingly important role in energy supply across various regions. However, as their scale and scope continue to expand, effective on-site monitoring to ensure their safety and efficiency has become a pressing issue. To ensure the normal operation of electrochemical energy storage power stations, a series of measures are needed to monitor and manage these devices. First, advanced sensors and instruments are required to measure key parameters of the power station, such as temperature, pressure, and power output. These sensors and instruments need regular calibration and maintenance to ensure accurate data acquisition and measurement. Second, a comprehensive monitoring system needs to be established to monitor the operating status of electrochemical energy storage power stations in real time. These systems need to have early warning and alarm functions, capable of issuing timely alerts in case of abnormalities so that personnel can take swift action.

[0003] However, the current monitoring system is based solely on numerical data and cannot monitor the health status of each energy storage unit in the power plant in real time, thus failing to conduct effective on-site monitoring in real time. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention proposes a method, system, equipment, and medium for on-site monitoring of electrochemical energy storage power stations.

[0005] The technical solution of the present invention is as follows:

[0006] On the one hand, the present invention provides a method for on-site monitoring of an electrochemical energy storage power station, comprising the following steps:

[0007] The historical operating parameters of each energy storage unit in the target chemical energy storage power station are obtained, preprocessed, and put into the historical operating dataset.

[0008] The K-Means clustering algorithm is used to cluster the data in the historical operation dataset to determine the optimal number of clusters K, resulting in K clustering results. Based on the operation parameter characteristics of the data in each clustering result, the energy storage unit state corresponding to each clustering result is determined. The energy storage unit state includes a healthy state and an unhealthy state, and the unhealthy state includes several abnormal sub-states.

[0009] A first state label is added to each clustering result to form a first training sample set, wherein the first state label is a healthy state and an unhealthy state; a second state label is added to the data in each clustering result belonging to the unhealthy state to form a second training sample set, wherein the second state label includes each abnormal sub-state;

[0010] A first neural network is trained using a first training sample set to obtain a first monitoring model for monitoring whether an energy storage unit is healthy; a second neural network is trained using a second training sample set to obtain a second monitoring model for identifying which abnormal sub-state the energy storage unit belongs to.

[0011] The real-time operating parameters of each energy storage unit in the target chemical energy storage power station are obtained, preprocessed, and then input into the first monitoring model to determine whether each energy storage unit is healthy. When any energy storage unit is detected to be in an unhealthy state, its corresponding data is input into the second monitoring model to identify which abnormal sub-state the energy storage unit is currently in.

[0012] In a preferred embodiment, the step of obtaining and preprocessing the historical operating parameters of each energy storage unit in the target chemical energy storage power station includes data cleaning and standardization. Specifically, data cleaning includes removing outliers, filling in missing values, and correcting erroneous values.

[0013] The standardization process involves the following steps:

[0014] The historical operating parameters of each energy storage unit are standardized as follows:

[0015]

[0016]

[0017] Where: x iN x′ is the Nth parameter in the i-th historical running parameter; iN It is the Nth parameter in the standardized i-th historical running parameter; s is the mean of the Nth parameter; k Let n be the standard deviation of the Nth parameter; n is the number of historical operating parameters; N is the number of parameters in each historical operating parameter.

[0018] Normalize the standardized historical operating parameters:

[0019]

[0020] in: To select the parameter with the smallest value among the historical operating parameters, The parameter with the largest value among the historical operating parameters is selected.

[0021] In a preferred embodiment, in the step of using the K-Means clustering algorithm to cluster the data in the historical dataset and determining the optimal number of clusters K, the method for determining the optimal number of clusters K is specifically as follows:

[0022] The optimal number of clusters K is found using the spotted hyena algorithm, as follows:

[0023] The spotted hyena population is initialized, with each spotted hyena having a unique K value, and K being greater than 3;

[0024] The fitness value of each spotted hyena individual is calculated based on the historical data set, where the fitness calculation formula is:

[0025]

[0026] Among them, D B (M) represents the sum of the inter-center centroid distances of all clustering results, and D... W (M) represents the sum of the intra-cluster distances of all clustering results;

[0027] The optimal spotted hyena individual in the current iteration is determined based on the fitness value, and the position of each spotted hyena individual is updated based on this optimal spotted hyena individual;

[0028] Repeat the above steps until the iteration ends, obtain the globally optimal spotted hyena individual, and determine the optimal number of clusters based on its corresponding K value.

[0029] In a preferred embodiment, the step of finding the optimal number of clusters K using the spotted hyena algorithm further includes an optimization step for the spotted hyena algorithm, specifically:

[0030] A competitive selection strategy is introduced in the step of determining the optimal spotted hyena individual for the current iteration based on its fitness value, as follows:

[0031] The competition process involves randomly selecting several spotted hyena individuals from the current population to form a competition group, and then selecting the spotted hyena individual with the best fitness value from the competition group to join the final group.

[0032] Repeat the above competition process until the spotted hyena individuals in the final group reach the preset threshold, then end the competition and select the spotted hyena individual with the best fitness value from the final group as the best spotted hyena individual in the current iteration.

[0033] In the hunting process of the spotted hyena algorithm, the prey's position is replaced by the optimal position of the spotted hyena, represented as:

[0034]

[0035]

[0036]

[0037] Where: m is the number of spotted hyenas; d represents the dimension of the population; This represents the position information of the i-th spotted hyena in the j-th dimension; t is the current iteration number; This represents the distance between the spotted hyena and its prey in the t-th iteration; This represents the position matrix of the prey in the t-th iteration. This represents the position matrix of the spotted hyena in the t-th iteration; B and E are both coefficient vectors, specifically expressed as:

[0038] B = 2·r1;

[0039] E = 2h·r² - h;

[0040] Where: t is the current iteration number; T is the maximum iteration number; h is the convergence factor; r1 and r2 are random numbers in [0, 1], and r1 < r2;

[0041] Introducing a chaotic sequence to control the convergence factor h, specifically:

[0042] The following chaotic sequence is generated using the initial parameters:

[0043]

[0044] Where γ is the independent variable that changes with time, and r c The preset length coefficient;

[0045] The chaotic sequence is standardized using a normalization method:

[0046]

[0047] A(t) = N m (t)·β(γ);

[0048] Wherein: T max This represents the maximum number of iterations for the spotted hyena algorithm. These are the upper and lower bounds of the function, respectively; t is the current iteration number; A(t) is the standardized sequence.

[0049] The convergence factor h at any number of iterations t is:

[0050] h = 5 - (t * (5 / T) max )+A(t).

[0051] On the other hand, the present invention also provides an on-site monitoring system for an electrochemical energy storage power station, comprising:

[0052] The dataset construction module is used to obtain the historical operating parameters of each energy storage unit in the target chemical energy storage power station, preprocess them, and put them into the historical operating dataset.

[0053] The clustering module uses the K-Means clustering algorithm to cluster the data in the historical operating dataset, determines the optimal number of clusters K, and obtains K clustering results. Based on the operating parameter characteristics of the data in each clustering result, the energy storage unit state corresponding to each clustering result is determined. The energy storage unit state includes a healthy state and an unhealthy state, and the unhealthy state includes several abnormal sub-states.

[0054] The training set construction module is used to add a first state label to each clustering result to form a first training sample set, wherein the first state label is a healthy state and an unhealthy state; and to add a second state label to the data in each clustering result belonging to the unhealthy state to form a second training sample set, wherein the second state label includes each abnormal sub-state;

[0055] The model training module is used to train a first neural network using a first training sample set to obtain a first monitoring model for monitoring whether the energy storage unit is healthy; and to train a second neural network using a second training sample set to obtain a second monitoring model for identifying which abnormal sub-state the energy storage unit belongs to.

[0056] The on-site monitoring module is used to acquire the real-time operating parameters of each energy storage unit in the target chemical energy storage power station, and input them into the first monitoring model after preprocessing to determine whether each energy storage unit is healthy. When any energy storage unit is detected to be in an unhealthy state, its corresponding data is input into the second monitoring model to identify which abnormal sub-state the energy storage unit is currently in.

[0057] In a preferred embodiment, the dataset construction module includes a data cleaning unit and a standardization processing unit;

[0058] The data cleaning unit is specifically used to clean the historical operating parameters of each energy storage unit, including removing outliers, filling in missing values, and correcting erroneous values.

[0059] The standardization processing unit is specifically used to standardize the historical operating parameters of each energy storage unit, as follows:

[0060]

[0061]

[0062] Where: x iN x is the Nth parameter in the i-th historical running parameter; i ′ N It is the Nth parameter in the standardized i-th historical running parameter; s is the mean of the Nth parameter; k Let n be the standard deviation of the Nth parameter; n is the number of historical operating parameters; N is the number of parameters in each historical operating parameter.

[0063] Normalize the standardized historical operating parameters:

[0064]

[0065] in: To select the parameter with the smallest value among the historical operating parameters, The parameter with the largest value among the historical operating parameters is selected.

[0066] In a preferred embodiment, the clustering module includes a spotted hyena algorithm unit for determining the optimal number of clusters K, specifically using the following method:

[0067] The optimal number of clusters K is found using the spotted hyena algorithm, as follows:

[0068] The spotted hyena population is initialized, with each spotted hyena having a unique K value, and K being greater than 3;

[0069] The fitness value of each spotted hyena individual is calculated based on the historical data set, where the fitness calculation formula is:

[0070]

[0071] Among them, D B (M) represents the sum of the inter-center centroid distances of all clustering results, and D... W (M) represents the sum of the intra-cluster distances of all clustering results;

[0072] The optimal spotted hyena individual in the current iteration is determined based on the fitness value, and the position of each spotted hyena individual is updated based on this optimal spotted hyena individual;

[0073] Repeat the above steps until the iteration ends, obtain the globally optimal spotted hyena individual, and determine the optimal number of clusters based on its corresponding K value.

[0074] In a preferred embodiment, the clustering module further includes an optimization unit for optimizing the spotted hyena algorithm. The specific optimization process is as follows:

[0075] A competitive selection strategy is introduced in the step of determining the optimal spotted hyena individual for the current iteration based on its fitness value, as follows:

[0076] The competition process involves randomly selecting several spotted hyena individuals from the current population to form a competition group, and then selecting the spotted hyena individual with the best fitness value from the competition group to join the final group.

[0077] Repeat the above competition process until the spotted hyena individuals in the final group reach the preset threshold, then end the competition and select the spotted hyena individual with the best fitness value from the final group as the best spotted hyena individual in the current iteration.

[0078] In the hunting process of the spotted hyena algorithm, the prey's position is replaced by the optimal position of the spotted hyena, represented as:

[0079]

[0080]

[0081]

[0082] Where: m is the number of spotted hyenas; d represents the dimension of the population; This represents the position information of the i-th spotted hyena in the j-th dimension; t is the current iteration number; This represents the distance between the spotted hyena and its prey in the t-th iteration; This represents the position matrix of the prey in the t-th iteration. This represents the position matrix of the spotted hyena in the t-th iteration; B and E are both coefficient vectors, specifically expressed as:

[0083] B = 2·r1;

[0084] E = 2h·r² - h;

[0085] Where: t is the current iteration number; T is the maximum iteration number; h is the convergence factor; r1 and r2 are random numbers in [0, 1], and r1 < r2;

[0086] Introducing a chaotic sequence to control the convergence factor h, specifically:

[0087] The following chaotic sequence is generated using the initial parameters:

[0088]

[0089] Where γ is the independent variable that changes with time, and r c The preset length coefficient;

[0090] The chaotic sequence is standardized using a normalization method:

[0091]

[0092] A(t) = N m (t)·β(γ);

[0093] Wherein: T max This represents the maximum number of iterations for the spotted hyena algorithm. These are the upper and lower bounds of the function, respectively; t is the current iteration number; A(t) is the standardized sequence.

[0094] The convergence factor h at any number of iterations t is:

[0095] h = 5 - (t * (5 / T) max )+A(t).

[0096] In another aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the on-site monitoring method for electrochemical energy storage power stations as described in any embodiment of the present invention.

[0097] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the on-site monitoring method for electrochemical energy storage power stations as described in any embodiment of the present invention.

[0098] The present invention has the following beneficial effects:

[0099] 1. This invention determines the number of states of an energy storage unit through clustering and performs state analysis and classification based on the clustering results, which can accurately determine the health status category of the energy storage unit. Two monitoring models are established according to whether the energy storage unit is healthy and what abnormal sub-state it is in. During on-site monitoring, it is possible to first determine whether the energy storage unit is healthy and then identify the abnormal type if it is not healthy, which reduces the amount of computation and can better meet the needs of on-site monitoring.

[0100] 2. This invention optimizes the spotted hyena algorithm by randomly selecting individuals with excellent fitness values ​​using a competition-based selection strategy, eliminating the need to sort all fitness values. This strategy is not only simple and easy to implement but also has low computational complexity, allowing for parallel processing and reducing the likelihood of getting trapped in local optima. Furthermore, by combining the chaotic sequence control convergence factor h with an appropriate normalization function, the prey search process can be better controlled, enabling the algorithm to search for the optimal solution more effectively and enhancing its adaptability and search capability. Attached Figure Description

[0101] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention. Detailed Implementation

[0102] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0103] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.

[0104] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0105] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0106] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.

[0107] Example 1:

[0108] See Figure 1 This embodiment provides a method for on-site monitoring of an electrochemical energy storage power station, including the following steps:

[0109] S100: Obtain historical operating parameters of each energy storage unit in the target chemical energy storage power station and preprocess them, then put them into the historical operating dataset. Historical operating parameters include charge and discharge efficiency, cycle life, energy loss, charging time, discharging time, voltage, current, temperature, etc. Based on the operating parameters of each energy storage unit, the operating status of the energy storage unit can be evaluated, and the health status of the energy storage unit can be determined.

[0110] S200. The K-Means clustering algorithm is used to cluster the data in the historical operating dataset, determining the optimal number of clusters K, resulting in K clustering results. Based on the operating parameter characteristics of the data in each clustering result, the energy storage unit state corresponding to each clustering result is determined. The energy storage unit state includes healthy and unhealthy states. Unhealthy states include several abnormal sub-states, which may occur due to different reasons, such as battery aging, malfunction, overcharging, or undercharging. In an unhealthy state, the energy storage unit may exhibit some abnormal characteristics, such as excessively long charging time, excessively short discharging time, and decreased charging and discharging efficiency. These operating parameter characteristics can serve as important criteria for judging the energy storage unit state corresponding to each clustering result. Operating parameters such as voltage, current, and temperature can also be used to assist in the judgment.

[0111] S300. Add a first state label to each clustering result to form a first training sample set. The first state label is a healthy state and an unhealthy state. Add a second state label to the data in each clustering result that belongs to the unhealthy state to form a second training sample set. The second state label includes each abnormal sub-state.

[0112] S400. Train the first neural network using the first training sample set to obtain a first monitoring model for monitoring whether the energy storage unit is healthy; train the second neural network using the second training sample set to obtain a second monitoring model for identifying which abnormal sub-state the energy storage unit belongs to.

[0113] The first monitoring model can be used to monitor the health status of energy storage units in real time. When an energy storage unit is in an unhealthy state, the first monitoring model can promptly detect it and issue an alarm, reminding staff to carry out maintenance or replacement. The application of this monitoring model can greatly reduce the impact of energy storage unit failures on the entire electrochemical energy storage power station, improving the stability and reliability of the power station. The second monitoring model can be used to identify abnormal sub-states of energy storage units. For example, when an energy storage unit experiences abnormal states such as insufficient capacity, low charging efficiency, or low discharging efficiency, the second monitoring model can accurately identify and classify them, helping staff to better understand the operating status and performance of the energy storage unit. The application of this monitoring model can greatly shorten maintenance and replacement time, improving the service life and efficiency of energy storage units. In this embodiment, the first training sample set includes energy storage unit data corresponding to healthy and unhealthy states, and the second training sample set includes energy storage unit data corresponding to different abnormal sub-states. The first neural network and the second neural network can choose to use the same neural network model, such as a convolutional neural network (CNN), a recurrent neural network (RNN), or a long short-term memory network (LSTM).

[0114] S500: Acquire real-time operating parameters of each energy storage unit in the target chemical energy storage power station, preprocess them, and input them into the first monitoring model to determine the health status of each energy storage unit. When any energy storage unit is detected to be in an unhealthy state, its corresponding data is input into the second monitoring model to identify which abnormal sub-state the energy storage unit currently belongs to. For different abnormal sub-states, corresponding measures can be taken for repair and maintenance. For example, for battery aging problems, the battery can be replaced or the charging and discharging rate can be reduced; for fault problems, repairs or component replacements can be performed; for overcharging or undercharging problems, the charging strategy can be adjusted or the battery can be replaced, etc.

[0115] Based on the method in this embodiment, the number of states of the energy storage unit is first determined by clustering, and the state analysis and classification are performed according to the clustering results, which can accurately determine the health status category of the energy storage unit. Two monitoring models are established according to whether the energy storage unit is healthy or not, and what abnormal sub-state it is in. During on-site monitoring, it is possible to first determine whether the energy storage unit is healthy, and then identify the abnormal type if it is not healthy, which reduces the amount of computation and can better meet the needs of on-site monitoring.

[0116] In a preferred embodiment of this invention, the step of obtaining and preprocessing the historical operating parameters of each energy storage unit in the target chemical energy storage power station includes data cleaning and standardization. Specifically, the data cleaning process includes:

[0117] 1. Outlier Removal: During data collection and entry, outliers are inevitable, which can negatively impact the accuracy of data analysis. Therefore, the first step in data cleaning is to remove these outliers. Outlier identification is typically based on statistical principles, determining whether a value is outlier based on the data distribution. After removing outliers, the data distribution will be more reasonable, leading to more accurate data analysis.

[0118] 2. Imputing Missing Values: During data collection and entry, some data may be missing, which could lead to inaccurate data analysis. Therefore, the second step in data cleaning is to impute these missing values. There are many methods for imputing missing values, including using the mean, median, mode, etc., to replace missing values, or using methods such as regression analysis to predict missing values. By imputing missing values, data integrity is ensured, and the accuracy of data analysis is improved.

[0119] 3. Correcting Error Values: Error values ​​may occur during data collection and entry, which can mislead data analysis. Therefore, the third step in data cleaning is to correct these error values. Methods for correcting error values ​​include using logical reasoning and comparison to identify and correct them. By correcting error values, data accuracy is ensured, and the accuracy of data analysis is improved.

[0120] The standardization process involves the following steps:

[0121] The historical operating parameters of each energy storage unit are standardized as follows:

[0122]

[0123]

[0124] Where: x iN x is the Nth parameter in the i-th historical running parameter; i ′ N It is the Nth parameter in the standardized i-th historical running parameter; s is the mean of the Nth parameter; k Let n be the standard deviation of the Nth parameter; n is the number of historical operating parameters; N is the number of parameters in each historical operating parameter.

[0125] Normalize the standardized historical operating parameters:

[0126]

[0127] in: To select the parameter with the smallest value among the historical operating parameters, The parameter with the largest value among the historical operating parameters is selected.

[0128] Different data generally have different dimensions. In order to make parameters with different dimensions comparable, this embodiment performs appropriate transformations and normalization on the data, compressing the data to the interval [0,1].

[0129] In a preferred embodiment of this example, in the step of using the K-Means clustering algorithm to cluster the data in the historical dataset and determining the optimal number of clusters K, the method for determining the optimal number of clusters K is specifically as follows:

[0130] The optimal number of clusters K is found using the spotted hyena algorithm, as follows:

[0131] The spotted hyena population is initialized, with each hyena having a unique K value, and K being greater than 3; the spotted hyena population initialization matrix is ​​as follows:

[0132]

[0133] in: is the position matrix of the spotted hyena population after initialization; m is the number of spotted hyenas in the population; d is the dimension of the population; This provides the position information of the i-th spotted hyena in the j-th dimension.

[0134] The fitness value of each spotted hyena individual is calculated based on the historical data set, where the fitness calculation formula is:

[0135]

[0136] Among them, D B (M) represents the sum of the inter-center centroid distances of all clustering results, and D... W (M) represents the sum of the intra-cluster distances of all clustering results;

[0137] Specifically, if the historical dataset after clustering contains m data points, then the centroids of all the data points are taken.

[0138]

[0139] Where: m is the sample size; x i For the i-th data;

[0140] After clustering is completed, the clustering result C of the i-th cluster is... i There is m i If there are 100 data points, then the centroid of the cluster is defined. for:

[0141]

[0142] In the clustering space, the sum of the Euclidean distances from each cluster to the centroid of the entire dataset is defined as the sum of the inter-cluster centroid distances, D. B (M):

[0143]

[0144] M represents the number of clustering results.

[0145] In the clustering space, the sum of the Euclidean distances from each sample in each cluster to the cluster center is the intra-cluster distance. Therefore, the sum of the intra-cluster distances of all M clusters is defined as the sum of the intra-cluster distances D. W (M):

[0146]

[0147] The fitness matrix of the spotted hyena can be represented as:

[0148]

[0149] in: This is the fitness matrix for spotted hyenas; It is the fitness value of the i-th spotted hyena.

[0150] The optimal spotted hyena individual in the current iteration is determined based on the fitness value, and the position of each spotted hyena individual is updated based on this optimal spotted hyena individual;

[0151] Repeat the above steps until the iteration ends, obtain the globally optimal spotted hyena individual, and determine the optimal number of clusters based on its corresponding K value.

[0152] In a preferred embodiment, the step of finding the optimal number of clusters K using the spotted hyena algorithm further includes an optimization step for the spotted hyena algorithm, specifically:

[0153] A competitive selection strategy is introduced in the step of determining the optimal spotted hyena individual for the current iteration based on its fitness value, as follows:

[0154] The competition process involves randomly selecting several spotted hyena individuals from the current population to form a competition group, and then selecting the spotted hyena individual with the best fitness value from the competition group to join the final group.

[0155] Repeat the above competition process until the spotted hyena individuals in the final group reach the preset threshold, then end the competition and select the spotted hyena individual with the best fitness value from the final group as the best spotted hyena individual in the current iteration.

[0156] In the hunting process of the spotted hyena algorithm, the prey's position is replaced by the optimal position of the spotted hyena, represented as:

[0157]

[0158]

[0159]

[0160] Where: m is the number of spotted hyenas; d represents the dimension of the population; This represents the position information of the i-th spotted hyena in the j-th dimension; t is the current iteration number; This represents the distance between the spotted hyena and its prey in the t-th iteration; This represents the position matrix of the prey in the t-th iteration. This represents the position matrix of the spotted hyena in the t-th iteration; B and E are both coefficient vectors, specifically expressed as:

[0161] B = 2·r1;

[0162] E = 2h·r² - h;

[0163] Where: t is the current iteration number; T is the maximum iteration number; h is the convergence factor; r1 and r2 are random numbers in [0, 1], and r1 < r2;

[0164] Introducing a chaotic sequence to control the convergence factor h, specifically:

[0165] The following chaotic sequence is generated using the initial parameters:

[0166]

[0167] Where γ is the independent variable that changes with time, and r c The preset length coefficient;

[0168] The chaotic sequence is standardized using a normalization method:

[0169]

[0170] A(t) = N m (t)·β(γ);

[0171] Wherein: T max This represents the maximum number of iterations for the spotted hyena algorithm. These are the upper and lower bounds of the function, respectively; t is the current iteration number; A(t) is the standardized sequence.

[0172] The convergence factor h at any number of iterations t is:

[0173] h = 5 - (t * (5 / T) mak )+A(t).

[0174] The hunting process after the hunt is completed is represented as follows:

[0175]

[0176]

[0177] in: It is the distance between other spotted hyenas in the population and the spotted hyena in the best position after the t-th iteration; It is the optimal position of the spotted hyena after the t-th iteration; It represents the positions of other spotted hyenas in the population after the t-th iteration. Other spotted hyenas in the population will move to the position of the best spotted hyena. At the same time, the fitness value of individuals in the population in the current iteration is calculated, and the position of the individual with the best fitness value will replace the position of the best individual in the previous generation.

[0178] It is known that the values ​​of h and E change with the number of iterations. When |E|<1, the spotted hyena attacks the prey; otherwise, it does not attack the prey.

[0179] Based on the above implementation methods, this embodiment randomly selects individuals with excellent fitness values ​​using a competition-based selection strategy, without needing to sort all fitness values. This strategy is not only simple and easy to implement, but also has low computational complexity, can be processed in parallel, and is less prone to getting trapped in local optima. Furthermore, by combining the chaotic sequence control convergence factor h with an appropriate normalization function, the prey search process can be better controlled, enabling the algorithm to search for the optimal solution more effectively, thus enhancing the algorithm's adaptability and search capability.

[0180] Example 2:

[0181] This embodiment provides an on-site monitoring system for an electrochemical energy storage power station, including:

[0182] The dataset construction module is used to obtain and preprocess the historical operating parameters of each energy storage unit in the target chemical energy storage power station and put them into the historical operating dataset; this module is used to implement the function of step S100 in Example 1, and will not be described in detail here.

[0183] The clustering module uses the K-Means clustering algorithm to cluster the data in the historical operating dataset, determines the optimal number of clusters K, and obtains K clustering results. Based on the operating parameter characteristics of the data in each clustering result, it determines the energy storage unit status corresponding to each clustering result. The energy storage unit status includes healthy status and unhealthy status, and the unhealthy status includes several abnormal sub-states. This module is used to implement the function of step S200 in Embodiment 1, and will not be described again here.

[0184] The training set construction module is used to add a first state label to each clustering result to form a first training sample set, wherein the first state label is a healthy state and an unhealthy state; and to add a second state label to the data in each clustering result belonging to the unhealthy state to form a second training sample set, wherein the second state label includes each abnormal sub-state; this module is used to implement the function of step S300 in embodiment one, and will not be described again here;

[0185] The model training module is used to train a first neural network using a first training sample set to obtain a first monitoring model for monitoring whether the energy storage unit is healthy; and to train a second neural network using a second training sample set to obtain a second monitoring model for identifying which abnormal sub-state the energy storage unit belongs to; this module is used to implement the function of step S400 in Embodiment 1, and will not be described in detail here;

[0186] The on-site monitoring module is used to acquire the real-time operating parameters of each energy storage unit in the target chemical energy storage power station, preprocess them, and input them into the first monitoring model to determine whether each energy storage unit is healthy. When any energy storage unit is detected to be in an unhealthy state, its corresponding data is input into the second monitoring model to identify which abnormal sub-state the energy storage unit currently belongs to. This module is used to implement the function of step S500 in Embodiment 1, which will not be described in detail here.

[0187] In a preferred embodiment of this invention, the dataset construction module includes a data cleaning unit and a standardization processing unit;

[0188] The data cleaning unit is specifically used to clean the historical operating parameters of each energy storage unit, including removing outliers, filling in missing values, and correcting erroneous values.

[0189] The standardization processing unit is specifically used to standardize the historical operating parameters of each energy storage unit, as follows:

[0190]

[0191]

[0192] Where: x iN x′ is the Nth parameter in the i-th historical running parameter; iN It is the Nth parameter in the standardized i-th historical running parameter; s is the mean of the Nth parameter; k Let n be the standard deviation of the Nth parameter; n is the number of historical operating parameters; N is the number of parameters in each historical operating parameter.

[0193] Normalize the standardized historical operating parameters:

[0194]

[0195] in: To select the parameter with the smallest value among the historical operating parameters, The parameter with the largest value among the historical operating parameters is selected.

[0196] In a preferred embodiment of this invention, the clustering module includes a spotted hyena algorithm unit for determining the optimal number of clusters K. The specific method is as follows:

[0197] The optimal number of clusters K is found using the spotted hyena algorithm, as follows:

[0198] The spotted hyena population is initialized, with each spotted hyena having a unique K value, and K being greater than 3;

[0199] The fitness value of each spotted hyena individual is calculated based on the historical data set, where the fitness calculation formula is:

[0200]

[0201] Among them, D B (M) represents the sum of the inter-center centroid distances of all clustering results, and D... W (M) represents the sum of the intra-cluster distances of all clustering results;

[0202] The optimal spotted hyena individual in the current iteration is determined based on the fitness value, and the position of each spotted hyena individual is updated based on this optimal spotted hyena individual;

[0203] Repeat the above steps until the iteration ends, obtain the globally optimal spotted hyena individual, and determine the optimal number of clusters based on its corresponding K value.

[0204] In a preferred embodiment of this invention, the clustering module further includes an optimization unit for optimizing the spotted hyena algorithm. The specific optimization process is as follows:

[0205] A competitive selection strategy is introduced in the step of determining the optimal spotted hyena individual for the current iteration based on its fitness value, as follows:

[0206] The competition process involves randomly selecting several spotted hyena individuals from the current population to form a competition group, and then selecting the spotted hyena individual with the best fitness value from the competition group to join the final group.

[0207] Repeat the above competition process until the spotted hyena individuals in the final group reach the preset threshold, then end the competition and select the spotted hyena individual with the best fitness value from the final group as the best spotted hyena individual in the current iteration.

[0208] In the hunting process of the spotted hyena algorithm, the prey's position is replaced by the optimal position of the spotted hyena, represented as:

[0209]

[0210]

[0211]

[0212] Where: m is the number of spotted hyenas; d represents the dimension of the population; This represents the position information of the i-th spotted hyena in the j-th dimension; t is the current iteration number; This represents the distance between the spotted hyena and its prey in the t-th iteration; This represents the position matrix of the prey in the t-th iteration. This represents the position matrix of the spotted hyena in the t-th iteration; B and E are both coefficient vectors, specifically expressed as:

[0213] B = 2·r1;

[0214] E = 2h·r² - h;

[0215] Where: t is the current iteration number; T is the maximum iteration number; h is the convergence factor; r1 and r2 are random numbers in [0, 1], and r1 < r2;

[0216] Introducing a chaotic sequence to control the convergence factor h, specifically:

[0217] The following chaotic sequence is generated using the initial parameters:

[0218]

[0219] Where γ is the independent variable that changes with time, and r c The preset length coefficient;

[0220] The chaotic sequence is standardized using a normalization method:

[0221]

[0222] A(t) = N m (t)·β(γ);

[0223] Wherein: T max This represents the maximum number of iterations for the spotted hyena algorithm. These are the upper and lower bounds of the function, respectively; t is the current iteration number; A(t) is the standardized sequence.

[0224] The convergence factor h at any number of iterations t is:

[0225] h = 5 - (t * (5 / T) max )+A(t).

[0226] Example 3:

[0227] This embodiment proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the on-site monitoring method for electrochemical energy storage power stations as described in any embodiment of the present invention.

[0228] Example 4:

[0229] This embodiment proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the on-site monitoring method for electrochemical energy storage power stations as described in any embodiment of the present invention.

[0230] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0231] Those skilled in the art will recognize that the units and algorithm steps described in the embodiments disclosed herein can be implemented using electronic hardware, computer software, or a combination of electronic hardware and software. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0232] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0233] In the several embodiments provided in this application, any function, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0234] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for on-site monitoring of an electrochemical energy storage power station, characterized in that, Includes the following steps: The historical operating parameters of each energy storage unit in the target electrochemical energy storage power station are obtained, preprocessed, and put into the historical operating dataset. The K-Means clustering algorithm is used to cluster the data in the historical operation dataset to determine the optimal number of clusters K, resulting in K clustering results. Based on the operation parameter characteristics of the data in each clustering result, the energy storage unit state corresponding to each clustering result is determined. The energy storage unit state includes a healthy state and an unhealthy state, and the unhealthy state includes several abnormal sub-states. Add a first state label to the data in each clustering result to form a first training sample set. The first state label is a healthy state and an unhealthy state. A second state label is added to the data in each cluster result that belongs to the unhealthy state to form a second training sample set. The second state label includes each abnormal sub-state. A first neural network is trained using a first training sample set to obtain a first monitoring model for monitoring whether an energy storage unit is healthy; a second neural network is trained using a second training sample set to obtain a second monitoring model for identifying which abnormal sub-state the energy storage unit belongs to. The real-time operating parameters of each energy storage unit in the target electrochemical energy storage power station are obtained, preprocessed, and then input into the first monitoring model to determine whether each energy storage unit is healthy. When any energy storage unit is detected to be in an unhealthy state, its corresponding data is input into the second monitoring model to identify which abnormal sub-state the energy storage unit is currently in. In the step of using the K-Means clustering algorithm to cluster the data in the historical dataset and determine the optimal number of clusters K, the method for determining the optimal number of clusters K is as follows: The optimal number of clusters K is found using the spotted hyena algorithm, as follows: The spotted hyena population is initialized, with each spotted hyena having a unique K value, and K being greater than 3; The fitness value of each spotted hyena individual is calculated based on the historical data set, where the fitness calculation formula is: ； in, This is the sum of the inter-center centroid distances of all clustering results. This is the sum of the intra-cluster distances of all clustering results; The optimal spotted hyena individual in the current iteration is determined based on the fitness value, and the position of each spotted hyena individual is updated based on this optimal spotted hyena individual; Repeat the above steps until the iteration ends, obtain the globally optimal spotted hyena individual, and use its corresponding K value as the optimal number of clusters; The process of finding the optimal number of clusters K using the spotted hyena algorithm also includes an optimization step, specifically: A competitive selection strategy is introduced in the step of determining the optimal spotted hyena individual for the current iteration based on its fitness value, as follows: The competition process involves randomly selecting several spotted hyena individuals from the current population to form a competition group, and then selecting the spotted hyena individual with the best fitness value from the competition group to join the final group. Repeat the above competition process until the spotted hyena individuals in the final group reach the preset threshold, then end the competition and select the spotted hyena individual with the best fitness value from the final group as the best spotted hyena individual in the current iteration. In the hunting process of the spotted hyena algorithm, the prey's position is replaced by the optimal position of the spotted hyena, represented as: ； ； ； Where: m is the number of spotted hyenas; d represents the dimension of the population; This represents the position information of the i-th spotted hyena in the j-th dimension; t is the current iteration number; This represents the distance between the spotted hyena and its prey in the t-th iteration; This represents the position matrix of the prey in the t-th iteration. This represents the position matrix of the spotted hyena in the t-th iteration; B and E are both coefficient vectors, specifically expressed as: ； ； Where: h is the convergence factor; and It is a random number in the range [0, 1], and < ; Introducing a chaotic sequence to control the convergence factor h, specifically: The following chaotic sequence is generated using the initial parameters: in, Let be the independent variable that changes with time. The preset length coefficient; The chaotic sequence is standardized using a normalization method: ； ； in: This represents the maximum number of iterations for the spotted hyena algorithm. , These are the upper and lower limits of the function, respectively. This represents the current iteration number; For standardized sequences; At any number of iterations The lower convergence factor h is: 。 2. The method for on-site monitoring of an electrochemical energy storage power station according to claim 1, characterized in that, The step of obtaining and preprocessing the historical operating parameters of each energy storage unit in the target electrochemical energy storage power station includes data cleaning and standardization. Specifically, data cleaning includes removing outliers, filling in missing values, and correcting erroneous values. The standardization process involves the following steps: The historical operating parameters of each energy storage unit are standardized as follows: ； ； in: For the first The first of the historical operating parameters One parameter; For the standardized first The first of the historical operating parameters One parameter; For the first The mean of each parameter; For the first The standard deviation of each parameter; The number of historical operating parameters; The number of parameters in each historical running parameter; Normalize the standardized historical operating parameters: ； in: To select the parameter with the smallest value among the historical operating parameters, The parameter with the largest value among the historical operating parameters is selected.

3. A field monitoring system for an electrochemical energy storage power station, characterized in that, include: The dataset construction module is used to obtain the historical operating parameters of each energy storage unit in the target electrochemical energy storage power station, preprocess them, and put them into the historical operating dataset. The clustering module uses the K-Means clustering algorithm to cluster the data in the historical operating dataset, determines the optimal number of clusters K, and obtains K clustering results. Based on the operating parameter characteristics of the data in each clustering result, the energy storage unit state corresponding to each clustering result is determined. The energy storage unit state includes a healthy state and an unhealthy state, and the unhealthy state includes several abnormal sub-states. The training set construction module is used to add a first state label to the data in each clustering result to form a first training sample set, wherein the first state label is a healthy state and an unhealthy state. A second state label is added to the data in each cluster result that belongs to the unhealthy state to form a second training sample set. The second state label includes each abnormal sub-state. The model training module is used to train a first neural network using a first training sample set to obtain a first monitoring model for monitoring whether the energy storage unit is healthy; and to train a second neural network using a second training sample set to obtain a second monitoring model for identifying which abnormal sub-state the energy storage unit belongs to. The on-site monitoring module is used to acquire the real-time operating parameters of each energy storage unit in the target electrochemical energy storage power station, and input them into the first monitoring model after preprocessing to determine whether each energy storage unit is healthy. When any energy storage unit is detected to be in an unhealthy state, its corresponding data is input into the second monitoring model to identify which abnormal sub-state the energy storage unit is currently in. The clustering module includes a spotted hyena algorithm unit, used to determine the optimal number of clusters K, specifically using the following method: The optimal number of clusters K is found using the spotted hyena algorithm, as follows: The spotted hyena population is initialized, with each spotted hyena having a unique K value, and K being greater than 3; The fitness value of each spotted hyena individual is calculated based on the historical data set, where the fitness calculation formula is: ； in, This is the sum of the inter-center centroid distances of all clustering results. This is the sum of the intra-cluster distances of all clustering results; The optimal spotted hyena individual in the current iteration is determined based on the fitness value, and the position of each spotted hyena individual is updated based on this optimal spotted hyena individual; Repeat the above steps until the iteration ends, obtain the globally optimal spotted hyena individual, and use its corresponding K value as the optimal number of clusters; The clustering module also includes an optimization unit for optimizing the spotted hyena algorithm. The specific optimization process is as follows: A competitive selection strategy is introduced in the step of determining the optimal spotted hyena individual for the current iteration based on its fitness value, as follows: The competition process involves randomly selecting several spotted hyena individuals from the current population to form a competition group, and then selecting the spotted hyena individual with the best fitness value from the competition group to join the final group. Repeat the above competition process until the spotted hyena individuals in the final group reach the preset threshold, then end the competition and select the spotted hyena individual with the best fitness value from the final group as the best spotted hyena individual in the current iteration. In the hunting process of the spotted hyena algorithm, the prey's position is replaced by the optimal position of the spotted hyena, represented as: ； ； ； Where: m is the number of spotted hyenas; d represents the dimension of the population; This represents the position information of the i-th spotted hyena in the j-th dimension; t is the current iteration number; This represents the distance between the spotted hyena and its prey in the t-th iteration; This represents the position matrix of the prey in the t-th iteration. This represents the position matrix of the spotted hyena in the t-th iteration; B and E are both coefficient vectors, specifically expressed as: ； ； Where: t is the current iteration number; h is the convergence factor; and It is a random number in the range [0, 1], and < ; Introducing a chaotic sequence to control the convergence factor h, specifically: The following chaotic sequence is generated using the initial parameters: in, Let be the independent variable that changes with time. The preset length coefficient; The chaotic sequence is standardized using a normalization method: ； ； in: This represents the maximum number of iterations for the spotted hyena algorithm. , These are the upper and lower limits of the function, respectively. This represents the current iteration number; For standardized sequences; At any number of iterations The lower convergence factor h is: 。 4. The on-site monitoring system for an electrochemical energy storage power station according to claim 3, characterized in that, The dataset construction module includes a data cleaning unit and a standardization processing unit; The data cleaning unit is specifically used to clean the historical operating parameters of each energy storage unit, including removing outliers, filling in missing values, and correcting erroneous values. The standardization processing unit is specifically used to standardize the historical operating parameters of each energy storage unit, as follows: ； ； in: For the first The first of the historical operating parameters One parameter; For the standardized first The first of the historical operating parameters One parameter; For the first The mean of each parameter; For the first The standard deviation of each parameter; The number of historical operating parameters; The number of parameters in each historical running parameter; Normalize the standardized historical operating parameters: ； in: To select the parameter with the smallest value among the historical operating parameters, The parameter with the largest value among the historical operating parameters is selected.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the on-site monitoring method for electrochemical energy storage power stations as described in any one of claims 1 to 2.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the on-site monitoring method for electrochemical energy storage power stations as described in any one of claims 1 to 2.

Images